# ETT (ETT Group of Companies Inc.): full reference > ETT builds AI and automation into the work your teams already do, with > clear ownership, human oversight and a record of every decision the > system takes. Headquartered in Dallas, Texas, United States, with > offices in London, United Kingdom and Vancouver, British Columbia, Canada. This is the long form of https://ettgroup.ai/llms.txt, generated from the same data the site renders. The rules in /llms.txt under "How to read the figures on this site" apply to every figure below. ETT does not publish fees; scope, duration and commercial terms are agreed in writing per engagement. ## Platform relationship ETT Group of Companies Inc. is a UiPath Authorized Partner and a UiPath Services Certified Partner, and resells UiPath licences. Our delivery team holds UiPath technical certifications. ## Routes to outcome https://ettgroup.ai/routes-to-outcome ### Step 1: AI Executive Briefing https://ettgroup.ai/services/ai-executive-briefing - Question it answers: Where would AI actually pay back in our operation? - What it is: A working session for your leadership team: a reality check, the scoring method, and a live map of your own operation. - Duration: 90 minutes - Who is involved: Six to ten executives, with a 15-minute sponsor call beforehand - Leave with: A one-page opportunity map within 48 hours; Two or three candidate processes; An honest next step, including "nothing yet" - Decision at the end: Is there anything worth pursuing, and which route fits it? ### Step 2: Executive Value Day https://ettgroup.ai/services/executive-value-day - Question it answers: Which AI investments will our operations own, and pay back inside nine months? - What it is: One working day, with pre-work, that turns ambition into opportunity charters owned by the executives accountable for the work. - Duration: One day, plus pre-work - Who is involved: Eight to twelve executives, led by a CEO or COO sponsor - Leave with: An enterprise value map; One-page opportunity charters; A ranked investment shortlist and decisions log - Decision at the end: Which charters go forward, and who owns each one? ### Step 3: Business Process Workshop https://ettgroup.ai/services/business-process-workshop - Question it answers: What is this process really costing us, and what should we automate, fix or leave alone? - What it is: A fixed-scope workshop that maps the processes, sets the baseline and ranks what to automate, fix or leave alone. No system access needed. - Duration: Four or six weeks - Who is involved: A sponsor, and about an hour each with the people who run the processes - Leave with: Process maps and a heat map; A baseline and benefits workbook; A prioritised roadmap and executive report - Decision at the end: Build, fix, or stop. ### Step 3: Process Mining and Discovery https://ettgroup.ai/services/process-mining-and-discovery - Question it answers: How do our core processes really run, measured from the systems themselves? - What it is: The same decision as the workshop, made from system event data and desktop capture, on the UiPath platform the build will run on. - Duration: About 12 weeks - Who is involved: Read-only access to the systems in scope, plus process owners and nominated users - Leave with: Mined process flows and variants; An error and rework baseline; A ranked roadmap and a platform ready to build on - Decision at the end: Which processes to build first, on evidence. ### Step 4: Proof of Value https://ettgroup.ai/services/proof-of-value - Question it answers: Does it work on our process, against measures our finance team set first? - What it is: One priority process built on your systems and judged against success criteria signed before the build. - Duration: Three to four weeks - Who is involved: A process owner, a finance signatory and integration access - Leave with: A working build; A results workbook and an Evidence File; A go or no-go record and a production plan - Decision at the end: Go, go with conditions, or no-go. ### Step 5: Pilot build https://ettgroup.ai/services/agentic-ai-and-automation - Question it answers: Can we take it live and show a measured return? - What it is: The proven process built into production on UiPath, with robots, AI agents and people as one governed process, then a hypercare period. - Duration: Scoped from the proof of value, then four weeks of hypercare - Who is involved: System owners, a named process owner and change windows - Leave with: A live process with exception paths and runbooks; Returns measured at 30 days, 90 days and month nine - Decision at the end: Scale it, or stop if a checkpoint misses. If it cannot pay back inside nine months, it should not be built. ### Step 6: Agentic Managed Services https://ettgroup.ai/services/agentic-managed-services - Question it answers: Who runs it on a Monday morning, and can we show what it did? - What it is: Monitoring, break-fix, drift checks and change control for your robots and agents, with a monthly evidence report. - Duration: Monthly - Who is involved: A named owner on your side for each process - Leave with: A monthly evidence report; A quarterly improvement backlog - Decision at the end: What to improve, extend or retire next. ### Alongside: AI Governance and Assurance https://ettgroup.ai/services/ai-governance-and-assurance - Question it answers: Could we hand an auditor the record of what our AI did? - What it is: An AI inventory, a risk baseline and evidence mapped clause by clause, with a named governance lead and quarterly assurance of AI outputs. - Duration: Three to four week baseline, then monthly or quarterly - Who is involved: System and risk owners - Leave with: An AI inventory and risk baseline; A clause-by-clause evidence map; A gap register and 90-day roadmap - Decision at the end: What to fix first, and what to evidence next. ### Alongside: Analect: AI cost and accuracy https://ettgroup.ai/analect - Question it answers: What will our AI cost, and can we prove what it got right? - What it is: An AI Spend Forecast built from your own bills, and an Accuracy Evidence File that checks AI outputs claim by claim. - Duration: Three weeks each - Who is involved: Your AI bills and usage, or a sample of AI outputs and their sources - Leave with: A forecast and budget for each AI workload; An Evidence File of outputs checked against their sources - Decision at the end: Where to cut AI cost, and which outputs to trust. Q: Do we have to start at the beginning? A: No. Start at the route that matches what you already know. If you have a priority process and a baseline, start with a proof of value. If you have a long list of ideas and no order, start with a briefing or a Value Day. What we will not skip is the baseline, because without it nobody can say whether the work paid back. Q: Why don't you publish prices? A: Because a price read before the outcome is understood tends to end the conversation in the wrong place. Each route has its scope, duration, deliverables and fee agreed in writing before any work starts, and the scope is set by the outcome you want rather than a package. The briefing and the Value Day are offered at no charge. Q: Can we stop after any route? A: Yes. Every route ends in a decision, and stopping is one of the options. You keep what that route produced: the map, the charters, the baseline, the Evidence File. Q: Where does governance come in? A: From the first conversation. AI Governance and Assurance and Analect run alongside the path rather than at the end of it, so the record exists before anyone asks for it. Q: What does a full route look like in time? A: The Value Day is designed around this timeline: the Value Day in month zero, a Business Process Workshop in months one and two, a proof of value in month three, build and go-live in months four to six, and returns measured in months seven to nine. ## Engagements in detail ### AI Executive Briefing https://ettgroup.ai/services/ai-executive-briefing The AI Executive Briefing is a 90-minute working session for your leadership team. No vendor pitch. We look at where AI is paying back in businesses like yours and where it is not, apply the same scoring we use in our workshops, and build a first map of your own operation in the room. You leave with a one-page opportunity map and an honest next step, which can be "nothing yet". In one sentence: A 90-minute working session for your leadership team that maps where AI would pay back in your operation, and a one-page opportunity map within 48 hours. What the buyer receives: - A reality check on where AI is paying back in operations like yours, and where it is not - The four lenses and the value, feasibility and risk scoring we use in our workshops - A first-pass heat map of your operation, built in the room - Two or three candidate processes worth a closer look - A one-page opportunity map, sent within 48 hours - An honest next step, which can be to fix a process first or to do nothing yet What the buyer provides: - A 15-minute call with the executive sponsor beforehand - Six to ten executives in the room: typically COO, CFO, CIO and the heads of operations and risk - No documents, data or system access How success is measured: - Whether the leadership team leaves with a shared view of where AI would pay back first - Whether the candidate processes are specific enough to scope The briefing finds candidates. It does not size a saving, and we do not report one. What can hold it up: - The right executives in the room for the full 90 minutes - A sponsor who can take the next step, or decide not to After it goes live: You can stop after the briefing and keep the map. Where there is something worth pursuing, we describe the route that fits it. How the work runs: 1. Sponsor call: Fifteen minutes with the executive sponsor to agree who should be in the room and what they want from it. No other preparation is needed. 2. Reality check: Where AI is paying back in operations like yours, and where it is not. 3. Method: The four lenses and the scoring, applied to examples from your business. 4. Live map: A first-pass heat map of your operation, built in the session, with two or three candidate processes. 5. Follow-up: The one-page opportunity map within 48 hours, with the next step we would recommend, if any. Q: How long does it take? A: Ninety minutes, plus a 15-minute call with your sponsor beforehand. The opportunity map follows within 48 hours. Q: Is there a charge? A: No. The briefing is offered at no charge to leadership teams who want a straight view of where AI would pay back. Q: What do we need to prepare? A: Nothing beyond the sponsor call. No documents, data or system access. Q: Who runs it? A: George McKenna, ETT's Chairman and CTO, with an ETT delivery lead: the people who would scope and run the work, not an account manager. Q: What happens next? A: You can stop and keep the map. Where something is worth pursuing, we recommend the route that fits: an Executive Value Day, a Business Process Workshop, a process mining discovery or a proof of value. Q: What is an AI Executive Briefing? A: A 90-minute working session for a leadership team that maps where AI and automation would pay back in their own operation, and ends with a one-page opportunity map sent within 48 hours. Q: Who should attend? A: Six to ten executives: typically the COO, CFO and CIO with the heads of operations and risk. The sponsor joins a 15-minute call beforehand to agree the room. Q: How is it different from the Executive Value Day? A: The briefing finds the candidates in 90 minutes. The Executive Value Day is a full day with pre-work that turns candidates into opportunity charters your executives own, each tested against a nine-month return. ### Executive Value Day https://ettgroup.ai/services/executive-value-day The Executive Value Day is a full-day working session for your leadership team. The morning is ideation and value creation. The afternoon takes the best ideas back into the operations each executive is accountable for, and formalises them as one-page opportunity charters. Nothing goes on a charter that cannot pay back inside nine months. In one sentence: One working day, with pre-work, that turns your leadership team's AI ambition into ranked opportunity charters, each owned by an executive and tested against a nine-month return. What the buyer receives: - Executive pre-work workbooks, issued ten business days before the day - A first-pass enterprise value map, agreed with your sponsor five days before - One-page opportunity charters: problem, owner, measure, return case and nine-month test - A ranked investment shortlist - A decisions log of what was agreed and who owns it - Written follow-up within three business days What the buyer provides: - Eight to twelve attendees: the CEO or COO as sponsor, the CFO, the accountable executives and the CIO or CTO - Each attendee's workbook, about 45 to 60 minutes - A 30-minute sponsor call before the day How success is measured: - Every charter has a named owner and a measure - Every charter passes, or is parked by, the nine-month return test - The first charter is specific enough to scope a Business Process Workshop without rediscovery The Value Day sizes opportunities roughly enough to rank them. The baseline that proves a saving is set in the workshop that follows. What can hold it up: - The accountable executives attending the whole day - Workbooks returned on time, so the value map is built from your evidence After it goes live: Charters, shortlist and decisions log are yours to keep. If you go further, a Business Process Workshop is scoped to the first charter. How the work runs: 1. Pre-work: Executive workbooks out ten business days before, a first-pass value map five days before, and a 30-minute sponsor call. 2. Morning: value: The value map played back, how AI changes the economics, ideation and sizing. 3. Afternoon: ownership: Ideas taken back into each executive's operation and formalised as one-page opportunity charters. 4. Portfolio and commitments: The shortlist ranked, owners named, and the decisions recorded before anyone leaves. 5. Follow-up: Value map, charters, shortlist and decisions log within three business days, with an optional route into a Business Process Workshop scoped to the charter. Q: How long does it take? A: One working day, with pre-work: a workbook for each attendee ten business days before, and a sponsor call. Written follow-up arrives within three business days. Q: Is there a charge? A: No, for qualified sponsors. We agree with your sponsor beforehand that the room, the pre-work and the decisions at the end of the day are real. Q: Who needs to be there? A: Eight to twelve people: the sponsor, the CFO, the executives accountable for the operations in scope, and the CIO or CTO, for the whole day. Q: What do we keep? A: The value map, the charters, the investment shortlist and the decisions log, whether or not you work with us afterwards. Q: What happens next? A: If you go further, a Business Process Workshop is scoped to the first charter. A typical route runs from the Value Day to a workshop in months one and two, a proof of value in month three, build and go-live in months four to six, and returns measured in months seven to nine. Q: What is an Executive Value Day? A: A full-day working session for a leadership team that turns AI ambition into ranked, one-page opportunity charters, each owned by an executive and tested against a nine-month return. Q: Who should attend? A: Eight to twelve people: the CEO or COO as sponsor, the CFO, the executives accountable for the operations in scope, and the CIO or CTO. Q: What happens after the day? A: You receive the value map, the charters, the investment shortlist and the decisions log within three business days. If you want to go further, a Business Process Workshop is scoped to the first charter. Q: What is an opportunity charter? A: A one-page statement of a single opportunity: the problem, the accountable owner, how success is measured, the return case and whether it passes the nine-month test. ### Business Process Workshop https://ettgroup.ai/services/business-process-workshop The Business Process Workshop is a fixed-price, fixed-scope engagement that maps how a function's processes run today, measures what they cost, and ranks where automation, AI agents or a simple fix would pay back. It runs for four to six weeks and ends with a prioritised roadmap and a decision, not a slide deck. In one sentence: A fixed-price workshop that maps how a function's processes run, sets the baseline and ranks what to automate, fix or leave alone, ending in a decision. What the buyer receives: - Process inventory and heat map across the function - Current-state process maps for the priority processes - Baseline and benefits workbook with low, expected and high cases, and every assumption written down - Opportunity register, each candidate scored for value, feasibility, risk and governance - Prioritised roadmap, reissued within two business days of the workshop - Executive report and workshop pack within five business days What the buyer provides: - A sponsor who can make the call at the end - Time with the people who run the processes, about an hour each - Whatever metrics exist, however partial - No system access and no licences: the workshop works from interviews, observation and documents How success is measured: - Leadership can choose between options on the evidence supplied - Every finding traces back to an interview, a capture session or a document - The first build can start without another discovery phase A workshop establishes what a case is worth. It does not itself realise a saving, and we do not report one. What can hold it up: - Process owners being available within the four or six week window - A sponsor who sets the scoring weights at kick-off - A named decision-maker who can accept or reject the roadmap After it goes live: The workshop ends with the report and roadmap, which are yours to keep. Where ETT continues into a proof of value or build, the roadmap becomes the plan of record. Two editions, one fixed price each - Focus: One function, four weeks. Includes: 12 processes long-listed; 3 processes mapped in detail; 8 stakeholder interviews; Half-day workshop; Full deliverable set. - Enterprise: Up to three functions, six weeks. Includes: 25 processes long-listed; 6 processes mapped in detail; 16 stakeholder interviews; Full-day workshop; Board presentation of the findings. How the work runs: 1. Kick-off and long list: Week one. We agree the scope, the scoring weights and the people we need, and long-list the function's processes. 2. Interviews and capture: Weeks one to three. We sit with the people doing the work and record how it actually gets done, including the workarounds. 3. Baseline: Week three. Effort, volumes, error and rework rates, and cost per process, with every source tagged so it can be checked. 4. The workshop: Week four. Findings played back for challenge, priorities agreed, roadmap issued. The Enterprise edition adds a board presentation. 5. Decide: Build, fix, or stop. A proof of value on the first priority process typically runs three to four weeks. Q: How long does it take? A: Four weeks for one function (Focus), six weeks for up to three (Enterprise). The pace is set mostly by how quickly we can get time with the people who run the processes. Q: How much does it cost? A: It is a fixed price for each edition, agreed in writing before we start: half on signature and half on delivery. Travel and expenses are excluded. Q: What access do you need? A: Time with process owners and whatever metrics exist. No system access and no licences: the workshop works from interviews, observation and documents. Q: What do we keep? A: Everything: maps, workbook, register, roadmap and report. They are yours whether or not you work with us afterwards. Q: What happens next? A: Build, fix or stop. If you go ahead, a proof of value on the first priority process typically runs three to four weeks. Q: What is a business process workshop? A: A structured engagement that maps how work runs today, measures what it costs and ranks the improvement and automation opportunities, ending in an agreed roadmap. Ours is fixed price and fixed scope, so you know the cost and the end date before it starts. Q: How is this different from process mining? A: The workshop works from interviews, observation and documents. Process mining works from system event logs. The workshop is the faster, cheaper starting point; mining adds hard data where the processes run through systems. Many programmes use both. Q: Do we need UiPath or any other licence? A: No. The workshop is tool agnostic and needs no licences or system access. If the roadmap points to a platform, we say which one and why. Q: What if the answer is "don't automate"? A: Then that is the answer, with the evidence behind it. A process that should be fixed or left alone is a valid finding, and it saves you a failed project. Q: Who from ETT runs it? A: An engagement lead, a process analyst (two on Enterprise) and an automation and AI architect. You deal with the people doing the work. ### AI Strategy and Delivery https://ettgroup.ai/services/ai-strategy-and-delivery ETT helps you prioritise AI and automation opportunities against business value, delivery effort and risk. Build a practical investment case and a delivery plan around the processes that matter most. In one sentence: A prioritised view of where AI and automation are worth investing in, with the business case and delivery plan behind the recommendation. What the buyer receives: - Opportunity register covering the processes reviewed and the candidates found - Prioritisation matrix scoring value, delivery effort and risk - Baseline and benefits model, with the assumptions written down - Dependencies and risk review across data, systems and permissions - Phased roadmap with a recommended first delivery - Recommended operating ownership for the work once it is live What the buyer provides: - Business priorities and the outcomes leadership wants - Access to process owners for interviews or walkthroughs - Whatever workflow metrics already exist, however incomplete - A view of the current platform landscape - Known constraints: budget cycle, regulatory scope, in-flight change How success is measured: - Quality of the investment decision: whether leadership can choose between options on the evidence supplied - Feasibility of the selected use cases when tested against data, systems and permissions - Readiness to begin delivery without a further discovery phase Strategy work establishes what a case is worth. It does not itself realise a saving, and we do not report one. What can hold it up: - Process owners being available within the agreed window - Access to the systems in scope, or an accurate description where access is not possible - A named decision-maker who can accept or reject the recommendation After it goes live: Strategy engagements end with a handover. Where ETT continues into delivery, the roadmap becomes the plan of record and is reviewed at each stage gate. How the work runs: 1. Orient: We map the organisation's goals, existing systems, data landscape, workflows and risks — identifying where AI can create value and where foundations need strengthening first. 2. Prove: We pressure-test the highest-value opportunity on a real, scoped use case, with the evidence and ROI modelling needed to justify investment before scaling. 3. Govern: We define the governance model, risk controls, accountability and operating model needed to adopt AI safely and demonstrably. 4. Scale: We shape the roadmap, architecture and delivery approach — prioritised use cases, stakeholder alignment and the operating model needed to take AI from plan into live delivery across the organisation. 5. Compound: We embed measurement, continuous improvement and internal capability so AI keeps delivering value and the organisation increasingly owns the outcome. Q: How long does this take? A: It depends on how many processes are in scope and how quickly process owners can be reached. The availability of your people, not our analysis, is usually what sets the pace. We agree a window and a fixed scope before starting rather than running an open-ended review. Q: What drives the cost? A: The number of processes reviewed, the number of stakeholders to interview, and how much usable measurement already exists. Where a process has no data at all, establishing a baseline is work in its own right and we price it as such. Q: What access do you need? A: Time with process owners, whatever workflow metrics exist, and a view of the platform landscape. We do not need production data access for strategy work, and we would rather not have it. Q: Who owns the output? A: Ownership of the solution, its configuration and the documentation is set out in the engagement contract before work starts. Where a third-party platform forms part of the solution, that platform's own licensing terms apply to it, and we identify which components those are rather than leaving the boundary vague. Q: What happens next? A: A conversation about the specific process or decision you have in mind. If there is a workable opportunity we will describe a scoped first engagement; if there is not, we will say so and explain why. Requesting a session does not commit you to anything. Q: What is AI Strategy and Delivery? A: It is the process of identifying where AI can create value, building a practical roadmap, and supporting the delivery of AI initiatives into live business use. Q: When should an organisation use AI consultancy? A: When it knows AI could create value but needs help prioritising use cases, assessing feasibility, building a business case or moving from pilots into production. Q: What is an AI opportunity assessment? A: A review of business processes, data, systems, risks and operational challenges to identify where AI is most likely to create measurable value. Q: How does ETT help with AI implementation? A: By helping define the roadmap, architecture, delivery plan, governance model and operating structure needed to move AI initiatives into live environments. Q: What is an AI Centre of Excellence? A: A structure that helps an organisation manage AI knowledge, standards, governance, best practice and capability across teams. ### Process Mining and Discovery https://ettgroup.ai/services/process-mining-and-discovery Process mining rebuilds how a process actually runs from the event logs your systems already keep, including the variants, rework and delays nobody documents. ETT, a UiPath Services Certified Partner, combines it with task mining and structured interviews on UiPath to give you a measured baseline and a ranked list of what to automate, fix or leave alone. In one sentence: A measured picture of how your priority processes really run, from system event logs, desktop capture and interviews, with a ranked roadmap and a UiPath platform ready to build on. What the buyer receives: - Process mining findings pack: flows, variants, bottlenecks and conformance against the intended process - Current-state maps (L1 to L3) for priority processes, with the decision and exception logic - Error and rework baseline per process - Register of pain points, risks and control observations - Opportunity heat map ranked by value, feasibility and risk - A governed UiPath Automation Cloud tenant, where licensing is in scope - Business Process Workshop and a 12 to 24 month roadmap What the buyer provides: - Read-only access to event data from the systems in scope, or exports where there is no API - Security and data-handling sign-off before anything connects - Process owners for interviews, and nominated users for task capture - Masking rules for personal or sensitive fields How success is measured: - Share of real cases the mined model explains, per process - Error, rework and cycle-time baseline agreed with the process owners - Leadership can rank and fund the first builds without further discovery Discovery measures the current state. The saving is realised by the build that follows, and is reported against this baseline. What can hold it up: - Data access: usually the critical path, especially for third-party systems with no API agreement - Process owners and nominated users being available for interviews and capture - A decision-maker for the workshop at the end After it goes live: The UiPath tenant stood up for discovery stays in place as the foundation for the first builds, so nothing is re-platformed. How the work runs: 1. Access and data: Weeks one to three. Security sign-off, read-only connections or extracts, data quality checks, platform stood up. Many third-party systems have no API agreement, so we plan for exports from week one. 2. Mine and capture: Weeks two to eight. Event-log analysis per process, task capture with nominated users, and interviews from leadership down to the people doing the work. 3. Synthesis: Weeks six to ten. One validated picture per process, reconciled between the data and the people. 4. Workshop and roadmap: Weeks ten to twelve. Findings challenged, priorities agreed, roadmap and documentation issued. 5. Build: The first priority process moves straight into a build on the same UiPath platform. Q: How long does this take? A: A typical discovery runs about 12 weeks across three processes. Agreeing data access is usually the longest single step, so we start it in week one and run interviews while it is arranged. Q: What drives the cost? A: The number of processes and systems in scope, how the event data can be reached (live connection or exports), and the UiPath licensing the discovery needs, which is quoted as its own line, separate from our fee. Q: What access do you need? A: Read-only access to event data from the systems in scope, or exports where there is no API; nominated users for task capture; and time with the process owners. Nothing connects before your security and data-handling sign-off. Q: Who owns the platform and the findings? A: Ownership of the solution, its configuration and the documentation is set out in the engagement contract before work starts. Where a third-party platform forms part of the solution, that platform's own licensing terms apply to it, and we identify which components those are rather than leaving the boundary vague. Q: What happens next? A: A conversation about the specific process or decision you have in mind. If there is a workable opportunity we will describe a scoped first engagement; if there is not, we will say so and explain why. Requesting a session does not commit you to anything. Q: What is process mining? A: Process mining uses the event logs business systems record (who did what, to which case, when) to rebuild how a process actually runs. It shows every variant, where work waits, and where it loops back for rework. It is increasingly called process intelligence. Q: What is the difference between process mining and task mining? A: Process mining works from system data and shows the end-to-end flow across many cases. Task mining records what people do on their desktops, so it shows how individual steps are done, including work in spreadsheets and email that never reaches a system log. Together they cover both. Q: Do we need process mining before automating with AI agents? A: For any process that runs through your core systems with real volume or risk, yes. An agent pointed at an undocumented process automates its exceptions. Mining first tells you which variant to build for, and where an agent adds anything over a rules-based robot. Q: What data does process mining need? A: An event log per process: a case ID, an activity name and a timestamp for each step, usually with the user and a few attributes. Most ERP, CRM, finance and case management systems hold this. Where there is no API, exports work. Q: Can the data stay in Canada? A: UiPath Automation Cloud can host your tenant in its Canada region, and Process Mining runs there. Depending on the model, generative AI calls from a Canadian tenant may be processed in the US; UiPath does not store the data there. UiPath's AI Trust Layer lets you restrict which regions model calls can go to, and some models run inside the Canada region. We choose models and settings to fit your residency rules, and check them again at go-live. Q: Is task mining surveillance? A: It should not be, and we set it up so it is not. It records only the applications you approve, for users who have agreed to take part, with sensitive fields masked. We agree the rules with your HR, privacy and security leads before anything is installed. ### Proof of Value https://ettgroup.ai/services/proof-of-value A Proof of Value builds one priority process on your real systems and data in three to four weeks, and measures it against success criteria your finance team signs before anything is built. It ends with a go, go-with-conditions or no-go decision, an Evidence File, and a production plan you can price. No criteria, no build. In one sentence: One priority process built on your systems in three to four weeks, judged against success criteria finance signs before the build, ending in a go, go-with-conditions or no-go decision. What the buyer receives: - V1 Success criteria, signed before the build and not moved afterwards - V2 Working build on your systems and data - V3 Results workbook against each criterion - V4 Evidence File: what the automation did, under whose permission, and where a person decided - V5 Go or no-go record with the reasons - V6 Production plan with a fixed-price pilot build, if the answer is go What the buyer provides: - A priority process, usually from a workshop, a discovery or an opportunity charter - A process owner and a finance signatory for the criteria - Integration access issued through your own secure process - Representative real cases to run How success is measured: - The criteria agreed in V1, and nothing chosen afterwards - Exception rate and how many cases needed a person - Cycle time and cost per completed case against the baseline A proof of value measures a short run on real cases. The saving is confirmed in production, at 30 days, 90 days and month nine of the pilot. What can hold it up: - Integration credentials arriving in the first week - Criteria signed before the build starts: no criteria, no build After it goes live: Whatever the decision, the Evidence File and results workbook are yours. A go moves into a fixed-price pilot build on the same platform. How the work runs: 1. Agree the criteria: Measures, baseline and pass marks set with your process owner and signed by finance. 2. Build: The process built on your systems with real data, exception paths and human review included. 3. Run and measure: Real cases through the build, measured against the criteria, with the evidence recorded as it runs. 4. Decide: Go, go with conditions, or no-go, recorded with the reasons. 5. Plan production: If it is a go, a production plan and a fixed-price pilot build, tested against a nine-month return. Q: How long does it take? A: Three weeks for one integration. Four weeks where a second integration or an agentic step is needed. Q: How is it priced? A: As a fixed fee, agreed in writing before work starts and sized to the version you need. We do not publish a price, because the right scope depends on the process and the criteria. Q: What access do you need? A: Integration access through your own secure process, representative real cases, a process owner, and a finance signatory for the success criteria. Q: What do we keep? A: The criteria, the results workbook, the Evidence File and the decision record, whatever the answer. Q: What happens next? A: Go, go with conditions, or no-go. A go comes with a production plan and a fixed-price pilot build, measured at 30 days, 90 days and month nine. Q: What is a proof of value? A: A short, fixed-scope build of one priority process on your own systems, judged against success criteria agreed and signed before the build starts, ending in a go, go-with-conditions or no-go decision. Q: How is a proof of value different from a proof of concept? A: A proof of concept shows something can work. A proof of value shows whether it pays back on your process, against measures your finance team set in advance. Q: How long does it take? A: Three weeks for one integration. Four weeks where a second integration or an agentic step is needed. Q: What happens if it fails? A: You get a no-go record with the reasons, the results workbook and the Evidence File. That is a result: it stops you funding a build that would not have paid back. ### Agentic Automation on UiPath https://ettgroup.ai/services/agentic-ai-and-automation Agentic automation combines AI agents, which interpret information and choose the next step within limits you set, with software robots that carry out fixed steps exactly, and people who approve what matters. ETT, a UiPath Services Certified Partner, designs and builds it on UiPath, orchestrated end to end, with every action logged and every exception routed to a named person. In one sentence: A working automation for one agreed workflow, connected to your systems, with the exception handling and documentation needed to run it. What the buyer receives: - Target process design: which steps run as robots, which as agents, which stay with people, and why - Integration and permission map: systems, accounts and the actions each component may take - Working automation deployed to your UiPath environment - Exception and escalation paths, designed before the build - Agent evaluation set and results: the test cases each agent must pass before go-live, and its score - Test evidence against the acceptance criteria agreed at the start - Runbooks and handover, or a move into our managed service What the buyer provides: - Access to the process and the people who run it today - Representative test data - Integration credentials, issued through your own secure process - Named system owners who can approve connections - Acceptance criteria agreed before build starts How success is measured: - End-to-end cycle time against the pre-agreed baseline - Successful completion rate - Exception rate, and how many exceptions a person had to resolve - Rework: how often a completed item had to be corrected - Cost per completed transaction - Agent decisions overturned by a reviewer, as a share of all decisions Not every workflow needs an agent. Where a process is predictable and the rules are stable, deterministic automation is cheaper to build, cheaper to run and easier to assure. We use AI interpretation where the input varies or the next step depends on meaning, and say which is which in the design. What can hold it up: - Integration credentials arriving on time — usually the single largest cause of delay - A stable target process; a workflow being redesigned in parallel cannot be automated reliably - Somebody with the authority to sign off what the automation may do without a human After it goes live: Automations need an owner after go-live. We hand over runbooks and monitoring, or run the work under our Agentic Managed Services agreement. Ongoing support is a separate commercial arrangement, not something included by default. How the work runs: 1. Orient: We find where automation creates measurable value, often from a workshop or process mining, and set the baseline. 2. Prove: We build one real process on UiPath and test it against criteria agreed before the build, with full traceability. 3. Govern: We set the limits, escalation paths and policies each robot and agent must respect, in the platform. 4. Scale: We roll the pattern out to live operations and adjacent processes, reusing modules rather than rebuilding. 5. Compound: We monitor, evaluate and improve, either handing over to your team or running it under our managed service. Q: How long does this take? A: A first workflow is a matter of weeks rather than months, but the schedule is set by integration access more than by build effort. Credentials arriving late is the most common cause of delay in this work, and we sequence around getting them early. Q: What drives the cost? A: The number of systems to integrate and the state of their interfaces; how much variation the process contains; and how much of the work genuinely needs AI interpretation rather than deterministic rules. A predictable process behind good APIs is considerably cheaper than a variable one behind a legacy screen. Q: What access do you need? A: Access to the process and the people who run it, representative test data, and integration credentials issued through your own secure process. We do not ask for shared accounts, and credentials are scoped to the actions the workflow needs. Q: Who owns what you build? A: Ownership of the solution, its configuration and the documentation is set out in the engagement contract before work starts. Where a third-party platform forms part of the solution, that platform's own licensing terms apply to it, and we identify which components those are rather than leaving the boundary vague. Q: What happens when it goes wrong? A: The workflow stops and escalates. Every automation we build has a defined failure path: what happens when a source is unavailable, a validation fails or permission is denied, and who it goes to. That path is designed at the start, not added after the first incident. Q: What is agentic automation? A: Agentic automation combines AI agents, software robots and people in one orchestrated process. Agents handle steps that need interpretation, robots handle fixed steps exactly, and people make the decisions that carry accountability. Everything runs within limits you set, and every action is recorded. Q: How is agentic automation different from RPA? A: RPA follows a fixed script and does the same thing every time, which makes it precise and easy to audit but brittle when inputs vary. An agent reasons about what to do next within the limits it is given. Most production processes use both: agents call robots for the precise steps. Q: When should we use an AI agent instead of a rules-based bot? A: When the next step depends on meaning rather than rules. If you can write the rule down and it holds, use a robot. It is cheaper and easier to assure. Q: How do you stop an AI agent doing something it shouldn't? A: The limits live in platform policy, not just in the agent's instructions. Each agent has a defined set of actions, a human escalation path, and a test set it must pass before go-live and after every change. Every action is logged against the permission it used. Q: Can you work with the UiPath robots we already have? A: Yes. Most agentic work builds on an existing estate rather than replacing it. We assess what you have, keep what works, and add agents where they remove a recurring failure. Q: Do we need to buy UiPath licences? A: Yes, for the UiPath components. As a UiPath reseller, we quote the licensing as its own line, separate from our fee and sized to what the roadmap needs rather than the whole platform. If you already have a UiPath agreement, we can work under it. Q: Is ETT a UiPath partner? A: Yes. ETT Group of Companies Inc. is a UiPath Authorized Partner and a UiPath Services Certified Partner, and resells UiPath licences. Our delivery team holds UiPath technical certifications. One team scopes the work, supplies the licences, builds on the platform and supports it afterwards. ### Conversational AI and Voice Automation https://ettgroup.ai/services/conversational-ai-and-voice-automation Connect voice and chat automation to your customer systems, knowledge and service workflows. ETT designs the interaction, the action behind it and the route to a person when judgement is needed. In one sentence: A voice or chat journey that resolves the routine requests you agree it should, and hands the rest to a person with the context attached. What the buyer receives: - Intent and journey map based on real interaction samples - Conversation design, including what the assistant will not attempt - Selected integrations to the systems that resolve the request - Human handover design: when it triggers, and what the agent receives - Testing and evaluation plan, run against held-back samples - Service performance reporting your operations team can read What the buyer provides: - Representative interaction types, ideally real transcripts or recordings - Approved knowledge sources the assistant may answer from - Escalation policy: what must always reach a person - Contact-centre architecture and telephony details - The languages and channels you actually need supported How success is measured: - Successful resolution: the request was completed, not merely contained - Repeat contact within a defined window - Customer satisfaction on handled interactions - Escalation quality: whether the agent received enough context to continue - Cost per resolved interaction Containment is one measure among these, not the definition of success. A contained call that produces a repeat contact tomorrow has resolved nothing. Supported languages, channels and integration scope are confirmed per deployment rather than promised in advance. What can hold it up: - Telephony and contact-centre change windows - Approval of the knowledge the assistant is allowed to answer from - Agreement on failure behaviour before launch: what callers hear when a connected system is unavailable After it goes live: Conversational systems drift as products and policies change. Plan for a review cadence and a named owner for the knowledge sources; we can run that review under a separate agreement. How the work runs: 1. Orient: We identify where voice or conversational AI can reduce friction, improve service journeys or create better visibility from customer interactions. 2. Prove: We prove the experience on a focused journey, measuring containment, resolution and customer outcomes before wider rollout. 3. Govern: We put consent, disclosure, recording, escalation routes and compliance controls in place so automated conversations are safe and accountable. 4. Scale: We implement voice agents and conversational AI into live environments, connecting them to the platforms and workflows they need across channels. 5. Compound: We monitor conversation performance, review escalation patterns, refine flows and analyse customer signals to keep improving over time. Q: How long does this take? A: The conversation design is rarely the constraint. Telephony change windows, contact-centre release cycles and knowledge approval usually set the timeline, and we sequence around them. Q: What drives the cost? A: The number of intents in scope, how many systems sit behind them, the languages and channels genuinely required, and the depth of testing. Adding a language is not a configuration toggle — it is design, testing and ongoing maintenance. Q: What access do you need? A: Representative interaction samples, ideally real transcripts or recordings; approved knowledge sources; your escalation policy; and details of the contact-centre architecture. Q: Who owns the design and the data? A: Ownership of the solution, its configuration and the documentation is set out in the engagement contract before work starts. Where a third-party platform forms part of the solution, that platform's own licensing terms apply to it, and we identify which components those are rather than leaving the boundary vague. Q: What happens when a customer gets stuck? A: They reach a person, with what has already happened attached. Handover design is part of the build, not an afterthought, and we measure whether the agent actually received enough context to continue. Q: What is conversational AI? A: Conversational AI allows people to interact with automated systems using natural language, by voice or chat. In an enterprise setting it can support customer service, internal queries, workflow routing and information retrieval. Q: What is a voice AI agent? A: An automated system that can understand spoken requests, respond naturally and, where appropriate, connect the conversation to systems or workflows. Q: How can conversational AI support contact centres? A: It can triage requests, answer repeatable queries, route calls, summarise interactions, support agents and trigger workflows across contact centre, CRM or ticketing systems. Q: Can conversational AI integrate with existing systems? A: Yes. It is most useful when it connects with CRM, contact centre platforms, ticketing tools, order management systems, scheduling tools or knowledge bases. Q: Does conversational AI replace human customer service teams? A: No. It should support teams by handling suitable repeatable interactions, improving routing, reducing manual effort and escalating more complex or sensitive cases to people. ### Agentic Managed Services https://ettgroup.ai/services/agentic-managed-services ETT's agentic managed service runs your automations and AI agents after go-live: monitoring, fixes when the systems around them change, drift checks on agent decisions, change control, and a monthly report of what ran, what failed and where people stepped in. It covers UiPath robots and agents, including estates built by someone else. In one sentence: Your automations and AI agents run, monitored and fixed after go-live, with drift checks, change control and a monthly evidence report. What the buyer receives: - Onboarding assessment of the estate: healthy, needs work, retire - Monitoring and incident handling against agreed service levels - Break-fix when the systems around an automation change - Agent evaluation re-run on a schedule and after every change - Versioned, tested and approved change control - Monthly evidence report and a quarterly improvement backlog What the buyer provides: - Access to the UiPath tenant and the systems the automations touch, under your accounts - A named owner for each process on your side - Agreed service levels and hours of cover - Your change and release windows How success is measured: - Job success and exception rates against the agreed service levels - Time to restore after a break - Agent decisions overturned by a reviewer, month on month - Value delivered against the original baseline The service keeps the operational record that frameworks such as OSFI E-23 and the EU AI Act ask for. It does not make you compliant on its own; that rests on your own control framework. What can hold it up: - Notice of changes to the systems the automations depend on - Credentials issued and rotated through your own secure process - An owner on your side who can approve changes After it goes live: Sized by the number of automations and agents in scope and the hours of cover you need, under a separate managed service agreement. Documentation stays current, so you can bring the work in-house. How the work runs: 1. Onboard: An assessment of the estate: what is healthy, what needs work and what should be retired, with the documentation brought up to date. 2. Baseline: Service levels, evaluation sets and the value baseline agreed, so the monthly report measures against something. 3. Run: Monitoring, break-fix, drift checks and change control, under your accounts and policies. 4. Report: A monthly evidence report your operations and risk teams can both use. 5. Improve: A ranked backlog of fixes and new candidates, reviewed with you every quarter. Q: What does it cost? A: It is sized by the number of automations and agents in scope and the hours of cover you need, and agreed under a separate managed service agreement. UiPath licensing, where we supply it, is its own line. Q: What hours do you cover? A: Cover hours are agreed per service, with extended cover available for critical processes. Q: What access do you need? A: Access to the UiPath tenant and the systems the automations touch, under your accounts and through your own secure credential process, plus a named owner on your side for each process. Q: Who owns the automations? A: You do. We run them under your accounts and your policies, and keep the documentation current, so you can bring the work in-house or move it elsewhere. Q: What happens next? A: A conversation about the specific process or decision you have in mind. If there is a workable opportunity we will describe a scoped first engagement; if there is not, we will say so and explain why. Requesting a session does not commit you to anything. Q: What is an agentic managed service? A: An ongoing service that runs, monitors and maintains automations and AI agents in production, so they keep working as the systems around them change, and you keep a record of what they did. Q: What is AgentOps? A: The practice of running AI agents in production: monitoring, evaluation, drift detection, change control and incident response. It does for agents what DevOps does for software. Q: What is agent drift, and how do you catch it? A: Drift is when an agent's decisions change over time without anyone changing the agent, because its inputs, tools or context did. We re-run a fixed set of test cases on a schedule and after every change, and compare the results with the baseline. Q: Do you only support automations ETT built? A: No. We take on existing UiPath estates after an onboarding assessment that tells you what is healthy, what needs work and what should be retired. Q: Who owns the automations? A: You do. We run them under your accounts and your policies, and keep the documentation current, so you can bring the work in-house or move it elsewhere. Q: How does this help with OSFI E-23 or the EU AI Act? A: It gives you the operational record those frameworks ask for: an inventory, monitoring, change history and evidence of human oversight, kept monthly rather than assembled the week before an audit. Your compliance position still rests on your own control framework. ### AI Governance and Assurance https://ettgroup.ai/services/ai-governance-and-assurance Named governance leadership for the AI you put into production. We baseline it against the frameworks your auditors, regulators and customers ask about, run the controls with you, and hand over evidence mapped clause by clause to ISO/IEC 42001, the EU AI Act and the NIST AI RMF. Three tiers, and each one stands alone. In one sentence: Named governance leadership for the AI you run, with an inventory, a risk baseline and evidence mapped clause by clause to ISO/IEC 42001, the EU AI Act and the NIST AI RMF. What the buyer receives: - AI system inventory with named owners - Risk baseline against ISO/IEC 27001 and 42001, SOC 2 and the OWASP Top 10 for LLM Applications - Clause-by-clause evidence map, each item rated partial, context or not evidenced - Gap register and a prioritised 90-day roadmap - A named governance lead, pre-release review and a quarterly risk committee report, on the retainer - A signed quarterly Assurance Pack of AI outputs checked claim by claim, at the top tier What the buyer provides: - Your list of AI systems in use and in build, however incomplete - Existing security policies and control framework - Access to system and risk owners - The questionnaires or regulatory obligations you need to answer How success is measured: - Coverage of the AI inventory against what is actually in use - Gaps closed against the 90-day roadmap - Evidence available when a reviewer asks for it We map evidence; we do not certify. ISO/IEC 42001 certification comes from an accredited body, and EU AI Act classification and role stay your decisions. What can hold it up: - Owners for each AI system who can answer for it - A risk owner who can accept a residual risk on the organisation's behalf After it goes live: Each tier stands alone. The baseline is yours to keep; the retainer and Assurance Pack run under a separate agreement. Three tiers. Each one stands alone. - AI Governance Baseline: Three to four weeks, fixed scope. Where you stand today, and what to fix first. Includes: AI system inventory with named owners; Risk baseline against ISO/IEC 27001 and 42001, SOC 2 and OWASP; Clause-by-clause evidence map with ratings; Gap register and a prioritised 90-day roadmap. - Governed Operations: Monthly. A named governance lead who runs the controls with you. Includes: Everything in the baseline, kept current; Named vCISO for AI; Pre-release review of every new use case and material change; Policy, oversight and escalation routes maintained; Incident support and a quarterly risk committee report. - Assurance at scale: Quarterly, per AI system. Governed Operations plus the Analect Accuracy Assurance Pack. Includes: Everything in Governed Operations; Up to 30 outputs a month certified against their sources; Reviewer decisions recorded on every flagged output; A signed quarterly pack (PDF, JSON and CSV) mapped to ISO/IEC 42001 and the EU AI Act. How the work runs: 1. Inventory: Every AI system in use and in build, with an owner. 2. Baseline: Risk assessed against the standards your buyers and regulators use. 3. Map: Evidence mapped clause by clause, with an honest rating for each. 4. Run: A named lead reviews new use cases and changes before release, and keeps the register current. 5. Assure: Outputs checked claim by claim each month, signed into a quarterly pack your risk committee can use. Q: How long does it take? A: The baseline takes three to four weeks. Governed Operations runs monthly, and the Assurance Pack runs a quarter at a time for each AI system. Q: How is it priced? A: The baseline is a fixed fee, the retainer is monthly and the Assurance Pack is quarterly per AI system, all agreed in writing before work starts. We do not publish prices, because scope depends on how many AI systems are in use and what you need to evidence. Q: What access do you need? A: Your list of AI systems, however incomplete, your existing policies and control framework, time with system and risk owners, and any questionnaire or regulatory request you are answering. Q: Do you certify us, or guarantee compliance? A: No. We map evidence to the clauses, rate it honestly and say what stays yours. Certification comes from an accredited body, and EU AI Act classification and role are your decisions. Q: What happens next? A: A conversation about the specific process or decision you have in mind. If there is a workable opportunity we will describe a scoped first engagement; if there is not, we will say so and explain why. Requesting a session does not commit you to anything. Q: Can ETT certify us to ISO/IEC 42001? A: No. Certification comes from an accredited certification body. We produce the evidence an auditor asks for, mapped to the relevant clauses and rated honestly, so the certification audit has something to test. Q: Does this make us EU AI Act compliant? A: Not on its own. Conformity is assessed per system, and your classification and role as provider or deployer are your decisions. We evidence the areas we can (human oversight, accuracy, monitoring and transparency design) and say what stays with you. Q: How does this help with OSFI Guideline E-23? A: E-23 asks federally regulated financial institutions to keep an inventory of models, rate their risk and monitor them across their life. The baseline creates the inventory and risk ratings for your AI systems, and the retainer keeps them current with a quarterly report to your risk committee. Q: What is the difference between this and vCISO for AI? A: vCISO for AI is the named leadership role. AI Governance and Assurance is the full service around it: the baseline, the evidence map, the retainer and, at the top tier, claim-by-claim assurance of AI outputs. ### vCISO for AI https://ettgroup.ai/services/vciso-for-ai ETT provides specialist security leadership for organisations introducing AI into business operations. Identify risks, define controls and give leadership a clear view of the decisions needed before and after deployment. In one sentence: Named security leadership for the AI work you are introducing, with the risks, controls and decisions written down and reviewed. What the buyer receives: - AI risk register covering the systems in scope - Governance responsibilities: who decides what, and who is accountable - Control priorities, sequenced against risk - Review of data and integration boundaries - Incident playbooks for the AI-specific failure modes - Executive reporting at an agreed cadence What the buyer provides: - The AI systems and use cases in scope - Existing security policy and control framework - Access to the technology and risk owners - Any regulatory obligations that apply to the deployment How success is measured: - Risks identified, then accepted, mitigated or transferred, with the decision recorded - Control coverage against the priorities agreed - Time to close the issues raised - Quality of the evidence available when a reviewer asks for it ETT provides advisory leadership and helps you reach readiness and alignment. We do not award certification, we do not guarantee compliance, and an advisory engagement does not include unlimited incident response. What can hold it up: - Access to the systems and the people who run them - A decision-maker who can accept a risk on the organisation's behalf - Agreement on scope: which AI systems the engagement covers, and which it does not After it goes live: The three support levels set out how much time and authority each engagement carries. Their exact review cadence, allocated advisory time, reporting and escalation rights are defined in the engagement letter before work begins. Levels of support - Foundational: Teams preparing for AI deployment or reviewing early-stage AI use. Includes: AI security risk assessment; Data classification and policy review; Regulatory gap analysis; Threat modelling for LLMs and voice AI; AI use policy guidance. - Operational: Teams operating AI, LLM or voice automation systems in production. Includes: Regular vCISO strategy sessions; API and integration security review; Prompt injection, manipulation and hallucination risk review; Incident response playbooks; Security scorecards and priority actions. - Leadership: Enterprise teams building AI into critical operations across multiple systems, regions or business units. Includes: Executive risk reporting; Board-level briefings; Data sovereignty advisory; Model update and rollback governance; Custom AI governance framework; Ongoing AI security roadmap. How the work runs: 1. Orient: We assess AI use cases, data flows, model behaviour, integrations, compliance exposure and current security maturity. 2. Prove: We validate the control approach against a real AI workload, confirming risks are understood and mitigations work before scale. 3. Govern: We define the governance model, policy framework, control requirements, escalation routes and security architecture for safe AI use, aligned to standards such as ISO/IEC 42001. 4. Scale: We support the implementation of controls, testing, monitoring and incident-response processes around AI systems moving into live operation. 5. Compound: We provide ongoing oversight, reporting, review and improvement so AI security keeps pace with changing models, workflows and risks. Q: How long does this take? A: Advisory engagements run over a period rather than to a delivery date. The support level sets the review cadence, the allocated advisory time and the reporting rhythm, and those are fixed in the engagement letter before work begins. Q: What drives the cost? A: The support level, the number of AI systems in scope, and the regulatory obligations that apply. An engagement covering one internal workflow is a different proposition from one covering a customer-facing deployment in a regulated market. Q: What access do you need? A: The AI systems and use cases in scope, your existing security policy and control framework, and time with the technology and risk owners. Q: Do you certify us, or guarantee compliance? A: No. We provide advisory leadership and help you reach readiness and alignment. Certification is awarded by an accredited body, not by ETT, and no advisory engagement can guarantee a compliance outcome. Any commitment on incident response is defined and bounded in the engagement letter rather than implied. Q: What happens next? A: A conversation about the specific process or decision you have in mind. If there is a workable opportunity we will describe a scoped first engagement; if there is not, we will say so and explain why. Requesting a session does not commit you to anything. Q: What is a vCISO for AI? A: A vCISO for AI provides specialist security and governance leadership for organisations deploying AI, LLMs, voice automation or agentic systems. The role helps assess risk, define controls and support safe adoption. Q: Why does AI need specialist security oversight? A: AI systems introduce risks around prompts, model outputs, sensitive data, integrations, automated workflows and governance. These often require controls beyond traditional cybersecurity measures. Q: What is prompt injection testing? A: Testing whether an AI system can be manipulated through malicious or unexpected instructions. It helps identify weaknesses in how the system follows prompts, accesses information or performs actions. Q: How does vCISO for AI support compliance? A: It can help review AI use against relevant requirements, policies and frameworks such as GDPR, HIPAA, PCI-DSS, SOC 2, ISO alignment and sector-specific obligations. Q: Does AI governance slow innovation down? A: Good governance should support adoption by creating clearer rules, controls and accountability. It helps organisations scale AI with more confidence rather than blocking progress. ### Data Discovery and Classification for AI https://ettgroup.ai/services/data-discovery-and-classification-for-ai ETT helps you locate relevant data, identify sensitive content and establish ownership and handling rules before it enters an AI workflow. In one sentence: A picture of what data exists across an agreed part of your estate, how sensitive it is, who owns it, and which sources are safe for an AI workflow to use. What the buyer receives: - Scoped data inventory across the repositories agreed at the start - Sensitivity labels applied against your classification policy - Ownership mapping for the data found - Exposure findings: over-permissioned, duplicated or misplaced content - Prioritised remediation plan - Approved source shortlist for AI use What the buyer provides: - The repositories in scope, agreed in writing - Read access, or a supervised scan window - Any existing classification policy - Responsible data owners for the areas covered How success is measured: - Coverage: how much of the agreed estate was actually reached - Classification quality, scored on a human-reviewed sample - Exposure identified - Remediation progress against the prioritised plan Scanning, classification, document extraction and answer retrieval are four different things. This service does the first two. It does not build the extraction or retrieval layer, and discovery tooling is not a document extraction engine. What can hold it up: - Scope agreed before work starts; an unscoped discovery has no completion point - Access to repositories, including any that require a change request - An owner for the remediation plan — findings without an owner do not get fixed After it goes live: Discovery is a point-in-time picture. Estates drift, so agree a re-scan cadence if the inventory needs to stay current. How the work runs: 1. Orient: We identify where data lives, how it is currently used, what systems are involved and where hidden or sensitive data may create risk. 2. Prove: We prove the approach on a defined data set, validating that discovery and classification surface the right risks and AI-ready sources. 3. Govern: We define the classification approach, data-handling rules, metadata structure and governance needed for safe AI use. 4. Scale: We apply discovery, classification and structuring across the agreed data sources, creating clearer visibility and more usable AI foundations. 5. Compound: We maintain data visibility over time, review classifications, update governance rules and improve readiness as AI use cases evolve. Q: How long does this take? A: It is set by the scope agreed at the start and by how long access approvals take. We will not run an unscoped discovery: without a boundary there is no point at which it is finished. Q: What drives the cost? A: The size and variety of the estate in scope, how many repository types are involved, and how much of the classification needs human review to be trustworthy. Q: What access do you need? A: Read access to the repositories in scope, or a supervised scan window where read access is not permitted, plus any existing classification policy and the responsible data owners. Q: Do you take our data off-site? A: No. Discovery and classification run against your estate under your controls. What leaves is the findings, not the content. Q: What happens next? A: A conversation about the specific process or decision you have in mind. If there is a workable opportunity we will describe a scoped first engagement; if there is not, we will say so and explain why. Requesting a session does not commit you to anything. Q: What is data discovery for AI? A: The process of locating and mapping the data an organisation holds, including structured and unstructured sources, so it can be assessed for AI readiness, governance and safe use. Q: What is AI data classification? A: Labelling data by sensitivity, compliance category, business function and potential AI use case. This helps organisations decide what data can be used, protected, restricted or excluded from AI workflows. Q: Why is data classification important before using AI? A: AI systems may access, retrieve or act on business data. Classification helps prevent sensitive, inaccurate or inappropriate data from being used in ways that create risk. Q: What types of data should be discovered and classified? A: Databases, documents, emails, transcripts, call logs, spreadsheets, customer records, internal knowledge bases, cloud app data and legacy system data. Q: How does data discovery support RAG and AI grounding? A: Discovery helps identify which sources can be used for retrieval-augmented generation and grounding. Classification and structuring then make those sources more reliable, traceable and suitable for AI use. ### Data and Analytics for AI https://ettgroup.ai/services/data-and-analytics-for-ai ETT connects and prepares operational data for reporting, retrieval and automation. Start with a specific decision or workflow, then build the data quality, access and context it needs. In one sentence: Approved information made usable for one named decision or workflow, with the quality, access and refresh rules to keep it that way. What the buyer receives: - Source and pipeline map for the data in scope - Data-quality baseline: completeness, freshness and known gaps - Agreed pipelines or retrieval layer - Dashboard, where the use case calls for one - Ownership and refresh rules for each source - Validation results against the agreed use case What the buyer provides: - Agreed sources and the permission model that governs them - Named data owners who can approve use - Stated quality expectations for the decision being supported - The reporting or retrieval use case the work is for How success is measured: - Data freshness against the agreed refresh target - Completeness against the fields the use case needs - Retrieval relevance, scored on a reviewed sample - Answer evaluation where the data feeds an AI workflow - Time spent preparing information, before and after This service makes approved information useful. Establishing what data exists and how it should be handled is Data Discovery and Classification — a different engagement, usually run first. What can hold it up: - Source system access and any vendor licensing needed to extract data - Data owners being available to confirm what a field actually means - A specific decision or workflow to build towards; open-ended data preparation has no measurable end After it goes live: Pipelines break when source systems change. Refresh rules, ownership and monitoring are part of the handover so that change has a named owner. How the work runs: 1. Orient: We assess the current data landscape, including systems, sources, quality issues, workflow dependencies and gaps that may limit AI performance. 2. Prove: We prove the data foundation against a real AI use case, validating that grounding, quality and pipelines deliver trustworthy outputs. 3. Govern: We define the data governance, quality controls and grounding approach needed for AI to work from trusted, well-managed information. 4. Scale: We implement the data architecture, pipelines, dashboards and AI-ready sources required to support live use cases at scale. 5. Compound: We monitor data quality, refine dashboards, update grounding sources and improve the intelligence layer as business needs change. Q: How long does this take? A: It is bounded by the use case, and we insist on having one. Data preparation without a specific decision to serve has no completion point and no way to tell whether it worked. Q: What drives the cost? A: The number of sources, the state they are in, and how much reconciliation the data needs before it can be trusted. Where the same field means different things in different systems, agreeing what it means is the work. Q: What access do you need? A: Agreed sources under your own permission model, the data owners who can confirm what a field represents, and the reporting or retrieval use case the work is for. Q: Who owns the pipelines? A: Ownership of the solution, its configuration and the documentation is set out in the engagement contract before work starts. Where a third-party platform forms part of the solution, that platform's own licensing terms apply to it, and we identify which components those are rather than leaving the boundary vague. Q: What happens next? A: A conversation about the specific process or decision you have in mind. If there is a workable opportunity we will describe a scoped first engagement; if there is not, we will say so and explain why. Requesting a session does not commit you to anything. Q: What does AI-ready data mean? A: Data that is clean, structured, relevant and accessible enough for AI systems to use reliably, with the context, quality and governance needed to support automation and decision-making. Q: What is AI grounding? A: Connecting AI systems to trusted business information so their responses and actions are based on relevant organisational data rather than generic model knowledge alone. Q: Why does data quality affect AI performance? A: Poor data can lead to inaccurate outputs, unreliable automation and weaker decision-making. AI systems need trusted, current and contextual data to perform effectively. Q: What types of data can support AI automation? A: Data from CRMs, ERPs, contact centre platforms, ticketing systems, spreadsheets, voice recordings, chat transcripts, documents and internal knowledge bases. Q: How can analytics support AI adoption? A: Analytics can show where processes are slowing down, where customer issues repeat, where demand is changing and where automation may create the most value. ### AI Infrastructure as a Service https://ettgroup.ai/services/ai-infrastructure-as-a-service ETT helps you assess and implement AI infrastructure against performance, cost, data location and operating requirements. Define what the workload needs before committing to capacity. In one sentence: A decision on where an AI workload should run, and the environment to run it in, sized against requirements you can actually state. What the buyer receives: - Workload requirements: model profile, expected demand, latency and data-location needs - Hosting options with a cost model for each - Architecture for the chosen option - Provisioning plan - Monitoring and support responsibilities, split between ETT, the provider and your team What the buyer provides: - Model and workload profile - Expected demand, including peaks - Latency requirements - Deployment constraints: data residency, network, procurement rules - Security requirements the environment has to satisfy How success is measured: - Workload performance against the agreed specification - Utilisation against provisioned capacity - Reliability over the measurement period - Cost against the modelled forecast Infrastructure is supplied by named providers; ETT designs, provisions and supports it. Available regions, current capacity, commercial unit and any minimum commitment are confirmed per engagement, because they change. What can hold it up: - A workload profile specific enough to size against; “we might need GPUs” is not one - Provider capacity in the region you need, at the time you need it - Your procurement and security review timelines After it goes live: Support boundaries are set per engagement: which party monitors what, what response is expected, and what falls to the underlying provider. How the work runs: 1. Orient: We assess the AI workload, model goals, data, performance and security requirements, and current infrastructure constraints. 2. Prove: We validate the architecture on a representative workload, confirming performance, cost and reliability before committing to scale. 3. Govern: We define the security, access, monitoring and operational controls the infrastructure must enforce for governed AI workloads. 4. Scale: We implement the infrastructure environment — compute, networking, storage, MLOps and integration — connecting the tooling and controls AI needs to run. 5. Compound: We review performance, capacity, cost and workload changes so the infrastructure continues to support AI use efficiently over time. Q: How long does this take? A: Design is quick. Provisioning depends on provider capacity in the region you need and on your own procurement and security review, both of which are worth starting early. Q: What drives the cost? A: The workload profile, expected demand including peaks, latency requirements and data-location constraints. Dedicated capacity costs more than contended capacity and earns that only for sustained workloads — we will tell you which yours is rather than assuming dedicated is always better. Q: What access do you need? A: A workload profile specific enough to size against: model, expected demand, latency requirements and any deployment constraints. Q: Who supplies the infrastructure? A: Named third-party providers supply the compute. ETT designs the environment, provisions it and supports it within the boundaries agreed per engagement. Available regions, current capacity, the commercial unit and any minimum commitment are confirmed at the time, because they change and a figure published on a web page would be out of date. Q: What happens next? A: A conversation about the specific process or decision you have in mind. If there is a workable opportunity we will describe a scoped first engagement; if there is not, we will say so and explain why. Requesting a session does not commit you to anything. Q: What is AI Infrastructure as a Service? A: It provides the compute, hosting, networking and operational environment needed to build, train, deploy or run AI workloads. Q: What is dedicated GPU hosting? A: Dedicated GPU hosting gives organisations access to GPU capacity for AI workloads such as model training, fine-tuning, inference or LLM experimentation, without relying solely on shared cloud instances. Q: When does a business need dedicated AI infrastructure? A: When workloads require predictable compute, stronger control, specific security requirements, data residency, MLOps capability or reduced reliance on generic shared cloud environments. Q: What is MLOps? A: The set of practices and tools used to deploy, monitor, manage and improve machine learning models in production environments. Q: Can AI infrastructure support RAG and vector databases? A: Yes. It can support RAG pipelines, vector databases, inference endpoints, model hosting and the wider tooling needed for AI applications. ## Agentic MSP ETT Group's managed services company. Agentic MSP runs back-office work for you with governed AI agents and a trained exception desk. Licensed decisions stay with your team, every outcome is recorded in an Evidence File, and the service is billed on verified outcomes rather than seats or hours. For independent insurance agencies and mid-market finance teams in the US and Canada. It has its own site, https://agentic-msp.com. ## Case studies in full Clients are described by sector rather than named. Vendor and platform names are real. Published 2026-09-11. ### Turning 220 siloed automations into a reusable framework https://ettgroup.ai/our-work/modular-automation-framework-global-logistics Client: a global logistics and supply chain operator. Sector: Logistics and supply chain. Region: UK and Europe. Engagement: Advisory and enablement. Duration: Phased programme, 4 to 6 week discovery then 12 to 18 months. Services: Automation strategy, Centre of Excellence design, Operating model, Agentic readiness. Technology: UiPath, RPA, OCR, Microsoft Power Platform, SAP, Oracle, BlueYonder. Consultancy engagement that moved a 220-automation estate from one-off builds to a governed library of reusable modules, cutting delivery time by 60 to 70 percent and ending duplicated support. The client: A global logistics and supply chain operator running a mature automation practice. The in-house team of around 30 people had built and now supported more than 220 automations across RPA, OCR and Microsoft Power Platform, spanning SAP, Oracle, BlueYonder, CRM and external partner systems. The challenge: The estate had grown one solution at a time. Every automation was a standalone build with its own integration code, its own error handling and its own support burden. Average delivery time for a new automation was around 30 days. When an upstream system changed, the same fix had to be applied to dozens of bots individually. Support effort was duplicated, delivery was slow, and the team had no clean route to introduce agentic capability on top of an estate that was not designed for reuse. What ETT did: We ran a consultancy-first engagement rather than a build contract. The client kept ownership of its automations and its team. Our job was to design the framework, prove it on live processes and enable the team to run it. - Discovery. Reviewed the full estate, shortlisted the 20 most used, most valuable and most reusable processes, and built a heatmap of shared integration touchpoints across SAP, Oracle, BlueYonder, CRM and external systems. - Framework and catalogue design. Defined module categories (inbound, processing, storage, outbound), created the first catalogue of reusable components with naming conventions and templates, wrote a requirements methodology aligned to reuse, and drafted the governance model covering module ownership and version control. - Pilot and operating model. Refactored high-value processes using the modular design, aligned support tickets to modules rather than individual bots, and defined KPIs for delivery speed, module reuse and support duplication. - Scale and enablement. Expanded the catalogue across the estate, retrofitted legacy bots, trained the 30-person team on module design and catalogue use, and stood up a Centre of Excellence with a continuous improvement process. - Agentic readiness. Identified where agentic automation complements the modular design, with a roadmap for self-healing modules, decision-support agents and autonomous orchestration. Commercially, the client chose between fixed cost per phase and a monthly retainer, so the engagement could flex with internal capacity. Delivered: - Discovery report and integration heatmap of common components - Modular automation framework and catalogue v1 - Requirements templates for business analysts - Pilot implementations proving the modular approach on live processes - Support alignment model mapping tickets to modules - Training and Centre of Excellence enablement pack for the in-house team - Agentic adoption roadmap with high-level proof-of-concept options Results: - Delivery time for new automations reduced by 60 to 70 percent against the previous 30-day average - Support aligned to modules rather than individual bots: fixes made once and applied across every automation that uses the module - Reusable module catalogue established and governed, with ownership and version control in place - Centre of Excellence stood up and the 30-person in-house team trained to run it - Agentic adoption roadmap agreed, with the estate prepared for self-healing modules and decision-support agents ### An AI assistant for social workers, kept entirely inside the platform https://ettgroup.ai/our-work/ai-case-file-assistant-social-care Client: a social care software provider. Sector: Social care and public sector technology. Region: UK. Engagement: Applied AI build. Duration: 3 month proof of concept, then phased roadmap. Services: Applied AI, Conversational AI, Regulated AI deployment, Private model hosting. Technology: Private cloud LLM deployment, Fine-tuned language model, Retrieval on regulatory corpus, SSO and MFA integration. A regulation-aware AI overlay that lets social workers question a case file in plain language, surfaces urgent risks, and never sends data outside the client's environment. The client: A UK software provider whose case management platform is used by fostering and adoption services to hold the full record for children in care: health plans, incident reports, risk assessments, placements and contact history. Their users are social workers under regulatory duty and time pressure. The challenge: Case files are large and grow continuously. Social workers were spending too much time searching for key details, summarising histories and checking that regulatory requirements were met. The bigger risk was what gets missed in that volume: an allergy buried in an old health plan, an incomplete risk assessment, a pattern of concern spread across several incident reports. The client wanted to help its users act faster and safer, but any solution had to process data entirely within their environment, respect existing user permissions, and produce answers that could be audited. What ETT did: We designed an AI assistant that sits as an overlay inside the existing case file view, behind the client's existing single sign-on and multi-factor authentication. It only sees the case file the user currently has open, and only what that user is permitted to see. The model is hosted within the client's environment, on-premises or private cloud. It is fine-tuned and grounded on the relevant regulatory framework, including the 2011 fostering regulations and national minimum standards, the platform's own data structures and representative case files, with prompt engineering focused on accuracy over fluency. Every interaction is logged for audit. The work is phased: - Phase 1, proof of concept (3 months). Contextual question and answer on the open case file (“what are the urgent matters for this child?”), a targeted set of regulatory checks and urgent risk surfacing (missing risk assessments, critical health concerns), data quality prompts for missing or inconsistent information, and a pilot with a selected group of social workers whose feedback shaped the next phase. - Phase 2, proactive analytics. Automatic scanning of case files for emerging risks and patterns such as repeated names, missed assessments and behavioural trends, with notifications to the responsible worker, and extension of the assistant to the foster carer and young person apps. - Phase 3, advanced features. Audio transcription and structured summarisation of recorded visits, cross-case pattern detection to spot common individuals or recurring incidents across multiple files, and integration with regulatory updates so the knowledge base stays current. Fixed-price build, with an annual support and maintenance agreement covering security updates, minor enhancements and regulatory compliance updates. Delivered: - In-platform AI assistant with contextual Q&A on the open case file - Regulation-grounded response layer aligned to fostering standards - Data quality prompts flagging missing or inconsistent records - Full interaction audit trail and role-based access enforcement - Private, on-platform model deployment with no external data processing - Pilot programme design and feedback loop with practising social workers - Phased roadmap for proactive analytics and cross-case pattern detection Results: - AI assistant delivered inside the case file view within the 3-month proof of concept and piloted with practising social workers - Contextual question and answer, regulatory checks and urgent risk surfacing in production use on live case files - All processing on-platform: no case data leaves the client's environment, and every interaction is logged for audit - Data quality prompts flagging missing risk assessments and inconsistent records at the point of entry - Roadmap for proactive analytics, foster carer and young person app extension, and cross-case pattern detection agreed for subsequent phases ### Evidence before automation for a financial services operator https://ettgroup.ai/our-work/process-discovery-automation-foundation-financial-services Client: a North American financial services company. Sector: Financial services. Region: Canada. Engagement: Discovery-led process mining and platform foundation. Duration: 12 week discovery programme, then build on the same platform. Services: Process mining, Task capture, Process discovery, Automation roadmap, Platform foundation, Governance. Technology: UiPath Process Mining, UiPath Task Mining, UiPath Task Capture, UiPath Automation Cloud. A won, fixed-fee engagement mapping three priority processes from system event data and from the people who run them, on a UiPath platform hosted in Canada that every later automation will run on. The client: A financial services company providing finance and insurance products distributed through a network of dealer partners across Canada. Operations span finance, HR, IT, compliance and dealer management, with quarterly rewards payments to partners as a high-volume, high-scrutiny process. The challenge: Leadership had a clear ambition: automate high-effort operational processes and build an automation capability that scales across the business. They chose three processes that between them touch HR and IT, finance and payments, and the dealer-facing commercial journey: employee onboarding, from signed contract to day-one readiness; quarterly rewards payments to dealer partners, the flagship; and dealer onboarding, from signed agreement through validation and activation. The risk was acting without visibility. Core processes had evolved over time, picking up manual workarounds, undocumented dependencies and inconsistent hand-offs between teams. Automating a process nobody fully understood would have locked in the inefficiency and multiplied the exceptions. Leadership also suspected error rates in the rewards process that had never been quantified. What ETT did: We are running the engagement discovery-led rather than workshop-led. Instead of asking stakeholders to describe their processes in a room, we observe how work actually flows through systems and people first, assemble the picture ourselves, and only then convene the business to validate and prioritise. - System-led process mining. UiPath Process Mining and Task Mining deployed against the in-scope systems to reconstruct end-to-end flows and variants from event data, surfacing bottlenecks, rework loops and process variants that are not documented anywhere. - Human-in-the-loop capture. Structured interviews with process owners across finance, HR, claims, compliance and operations, plus Task Capture of how work is actually completed (clicks, inputs, sequencing, workarounds), reconciled against the mined system view. This is also where we find processes that should exist and do not. - Synthesis. A single validated process landscape across in-scope functions, ownership and hand-off mapping, a consolidated register of pain points, risks and control observations, and a prioritised opportunity heatmap ranked by business impact, feasibility, risk reduction and effort saved. - Access planned around reality. Several third-party systems in scope have no API agreement. Rather than wait, interviews and task capture run first while data access is agreed system by system, using exports where there is no live connection. Access is the critical path in most discovery work, so it is planned in week one, not discovered in week three. - Error and rework baseline. Measured per process from the mined data and the capture sessions, with a specific line for incorrect payments in the rewards process. Every later business case rests on the client's own numbers rather than an estimate. - Business Process Workshop. Held once mapping and documentation are complete, not before. Findings played back to stakeholders for challenge and correction, priority processes confirmed, automation opportunities ranked. - Documentation and roadmap. ISO 9001-ready SOP-style documentation for selected processes, decision and exception logic mapping, systems touchpoint maps, effort baselines for ROI modelling, and a 12 to 24 month roadmap. In parallel we are standing up UiPath Automation Cloud, hosted in Canada, so prioritised opportunities move straight into build without re-platforming. Security and data-handling sign-off came before any system connection, and working files stay in the client's own Microsoft 365 environment. When the roadmap says build, a proof of value on the first priority process runs on the same platform in three to four weeks, rather than as a new project. Fixed professional services fee, with the UiPath licence as its own line. Delivered: - Executive summary report with leadership-level findings, themes and risks - Enterprise process landscape and visibility pack - Process mining findings pack with data-driven flows, variants and bottleneck analysis - Current-state process maps (L1 to L3) and ISO 9001-ready SOP documentation for priority processes - Task capture records with decision tree and exception logic mapping - Operational pain-point register and risk and control observations log - Effort, error and rework baselines for ROI modelling - Prioritised opportunity heatmap and 12 to 24 month automation roadmap - Deployed, governed UiPath Automation Cloud tenant, hosted in Canada, ready for the first builds Still in delivery. No results are published. Designed to deliver: - One validated picture of how each of the three processes really runs, from system data and the people doing the work - An error and rework baseline per process, including incorrect payments in the rewards process - Undocumented steps, side spreadsheets and missing processes surfaced, reducing key-person dependency - Automation opportunities ranked by value, feasibility and risk at a single Business Process Workshop - A governed UiPath Automation Cloud tenant, hosted in Canada, ready for the first build without re-platforming ### Email-to-case-system automation as a group-wide foundation https://ettgroup.ai/our-work/intelligent-document-processing-medico-legal-group Client: a medico-legal services group. Sector: Healthcare and legal services. Region: UK. Engagement: Intelligent document processing and automation foundation. Duration: 4 to 6 week proof of concept, 13 week production build. Services: Intelligent document processing, Process automation, Human-in-the-loop design, Platform foundation, Enablement. Technology: UiPath IXP, UiPath Document Understanding, UiPath AI Centre, UiPath Action Centre, UiPath Orchestrator, UiPath Automation Cloud, UiPath Automation Ops, Proclaim, Blue Prism. Automating the intake, extraction, validation and submission of instruction letters and medical reports into Proclaim, saving over £120k a year and built as the first module of a group-wide automation platform. The client: A UK group providing independent medical examinations, medico-legal reporting and related services through several acquired businesses, each historically running its own case management system, document workflows and automation tooling, including legacy Blue Prism. Group IT and automation governance was being consolidated centrally. The challenge: Thousands of instruction letters and medical reports arrive by email every year. Staff opened each one, read it, and re-keyed the content into the case management system by hand. Document formats varied from client to client. Checking that the right expert had reviewed the right documents meant cross-referencing multiple sources manually. When instructions were incomplete, someone had to chase the customer and escalate internally. The process was slow, error-prone, hard to scale as volumes grew (around 100 submissions a day in the first business unit alone), and it consumed skilled administrative capacity that should have been on case work and client service. Legacy RPA tooling and unsanctioned integration tools such as Zapier added cost and risk. What ETT did: We scoped the first use case, email-to-Proclaim automation, as the foundation for a group-wide UiPath platform rather than a one-off bot. Every component built for it (email handling, document parsing, validation logic, API integration) is reusable for the next automation. The solution is a set of coordinated agents on UiPath IXP: - Email monitoring watches dedicated inboxes and starts the workflow on arrival. - Data extraction uses OCR and NLP through UiPath Document Understanding and AI Centre to pull structured data from free-text emails and scanned attachments in varying layouts. - Data structuring maps extracted fields to the Proclaim schema and detects missing information. - Customer feedback sends automated follow-ups for incomplete submissions and reprocesses when the response arrives. - CMS integration submits to Proclaim via secure REST API with error handling and retry logic. - Escalation routes anything unresolved or low-confidence to UiPath Action Centre so staff stay in control of edge cases. A short proof of concept on UiPath Community licensing validated extraction quality and Proclaim integration on a live workflow before production licensing. The 13-week production build ran through requirements and design, Orchestrator and connector setup, extraction, the feedback loop, integration, escalation, testing and handover, with a named delivery team covering architecture, automation engineering, document understanding, integration, testing, training and support transition. Licensing was sized on the Unified Pricing platform tier with Platform Units for IXP, two unattended robots for the daily volume, and named users for development and exception handling. For the wider group, UiPath acts as the orchestration layer: existing Blue Prism bots are triggered via API or queue hand-off and migrated progressively rather than ripped out. Governance runs through UiPath Automation Ops. A structured enablement programme develops the group's own citizen (StudioX, Automation Hub) and professional (Studio, Action Centre) developers, with our support tapering to second line as internal capability grows. We flagged the use of Zapier as an information security risk requiring formal review. Delivered: - End-to-end email-to-Proclaim automation for instruction letters and medical reports - UiPath Document Understanding and AI Centre models for varied and scanned document formats - Automated customer follow-up loop for incomplete instructions - Action Centre exception queue and escalation workflows - Secure REST API integration with Proclaim, with error handling and retry - Hybrid orchestration model for existing Blue Prism assets - Runbooks, documentation, training and a support transition plan - Group-level pipeline of follow-on use cases across finance, QA, case triage, onboarding, customer service and compliance Results: - Over £120,000 in annual savings from the removal of 5 to 8 FTE equivalents of manual re-keying - Around 100 submissions a day processed automatically through UiPath IXP into Proclaim - Cases initiated in hours rather than days, with faster client turnaround and revenue recognition - £20,000 to £30,000 a year saved in rework and compliance cost through fewer data entry errors - Full audit trail on every submission, with exceptions handled through UiPath Action Centre - Payback inside 12 to 18 months, and a reusable platform for follow-on group use cases ### Taking the year-end squeeze out of finance audit and reconciliation https://ettgroup.ai/our-work/finance-audit-reconciliation-automation-investment-management Client: a UK investment and wealth management group. Sector: Financial services. Region: UK. Engagement: Intelligent automation build on existing platform. Duration: 8 weeks. Services: Intelligent document processing, Anomaly detection, Human-in-the-loop design, Finance automation, Performance dashboards. Technology: UiPath Automation Cloud, UiPath Document Understanding, UiPath AI Centre, UiPath Action Centre, UiPath Insights, Sage, Power BI. Extending an existing UiPath Automation Cloud platform into finance audit and year-end reconciliation with document extraction, anomaly detection, approval workflows and live audit KPIs, with no new licensing. The client: A UK investment and wealth management group with UiPath Automation Cloud already in production for investment platform data aggregation and portfolio rebalancing, delivered by us in an earlier phase. Unattended robots, Orchestrator, Automation Hub and Power BI dashboards were all in place. The challenge: Finance audit and year-end reconciliation was still manual: spreadsheets, email chains and fragmented data sources. Analysts pulled records from Sage by hand (10 to 12 hours a week each), matched receipts and invoices line by line, chased approvals through email and meetings, and compiled audit-ready documentation under tight deadlines. Every discrepancy needed manual investigation. Leadership had no real-time view of audit progress or bottlenecks. The pressure peaked at year-end, with extended hours, inconsistent documentation and a real retention risk in the finance team. What ETT did: We extended the platform the client already owned rather than sell them another one. No additional licensing was required. The build introduced four UiPath capabilities: - Document Understanding for receipts, invoices, remittance advice and financial documents, using pre-built models for billing, purchase orders and financial statements to accelerate deployment, with automatic retraining from data validated in Action Centre. - AI Centre machine learning skills to flag outliers, duplicates and potential fraud patterns in transaction data, trained on historical audit data. - Action Centre approvals so finance staff review low-confidence extractions, validate vendor name variations and invoice discrepancies, and approve reconciliations inside the workflow instead of over email. Unresolved anomalies escalate automatically. - Insights dashboards tracking time saved, error rates, audit readiness and cycle-time forecasts, with alerts for threshold breaches, building on the existing Power BI reporting. Pre-built audit reconciliation templates from UiPath's financial services library accelerated delivery. The programme ran over eight weeks: two weeks of discovery and design mapping Sage and the audit workflows, four weeks of build and test, two weeks of user acceptance and go-live. The design is horizontally scalable to monthly close, claims processing, regulatory reporting and treasury. Delivered: - Automated extraction and validation of receipts, invoices and supporting documents - AI Centre anomaly and duplicate detection on transaction data - Action Centre approval and exception workflows keeping finance in control - Real-time audit KPI dashboards in UiPath Insights for finance leadership - Reuse of the existing Automation Cloud tenant, robots and orchestration with no new licences - A pattern extendable to other finance processes Results: - Manual effort on audit and reconciliation reduced by up to 80 percent - Analyst data extraction time cut from around 12 hours a week to under 2 - Audit preparation reduced from around 3 weeks a year to under 1 - Total annual benefit of around £150,000 across time saved, error reduction, faster decision-making and strategic enablement - Payback in around 7.5 months, with year one savings exceeding delivery cost - No additional licensing: delivered entirely on the platform the client already owned ### From inbox to ERP without the re-keying https://ettgroup.ai/our-work/sales-order-processing-automation-manufacturing Client: a UK manufacturer and distributor of safety footwear and workwear. Sector: Manufacturing and distribution. Region: UK. Engagement: Intelligent document processing and ERP integration. Duration: 4 week proof of concept, 8 to 11 week production build. Services: Intelligent document processing, ERP integration, Human-in-the-loop design, Change and adoption, Managed automation. Technology: UiPath IXP, UiPath AI Centre, UiPath Action Centre, UiPath Orchestrator, UiPath Insights, OrderWise. Automating purchase order intake from customer emails and PDFs into OrderWise, with machine learning to resolve inconsistent product codes and an Action Centre queue for anything ambiguous. The client: A UK manufacturer and distributor of safety footwear and workwear, selling to trade customers who place orders by email, often as PDF purchase orders in their own formats, with operations spread across several units and an OrderWise ERP migration to cloud on the horizon. The challenge: Every order was keyed into OrderWise by hand from an email or attachment. Customer documents came in every format imaginable. Product codes were the hardest part: customers describe a product their own way, in their own shorthand for a model and a size, and someone had to interpret it and look up the internal SKU. Errors and delays followed, capacity was the constraint at peak, there was no structured way to handle exceptions and no visibility of throughput. Any mistake at intake became a delivery problem later. What ETT did: We started with a four-week fixed-cost proof of concept on the client's real purchase orders and SKU data, using UiPath Community and partner NFR licensing so there was no licence commitment until value was demonstrated. The proof of concept extracted data from sample orders with UiPath IXP, normalised SKU variants and simulated posting to OrderWise. The production build on UiPath Platform Standard automates the full flow: - Email listener monitors the Outlook orders inbox and triggers on arrival. - UiPath IXP performs generative document extraction on PDF attachments and email bodies for customer, delivery address, product codes and quantities, handling varied formats without template training. - SKU normalisation in UiPath AI Centre matches customer descriptions to internal codes using historical mappings, retraining on exceptions so accuracy improves over time. - Validation routes low-confidence extractions or ambiguous SKUs to Action Centre for human review, with an optional clarification email to the customer. - Order creation posts structured data to OrderWise via API with logging and error handling. - Monitoring in Orchestrator and Insights tracks processed orders, exceptions and performance for operational and executive reporting. Two further phases were scoped: enablement for an incoming internal IT hire (Studio, Orchestrator, AI Centre training) so the client owns the platform long term, and modularising the OrderWise integration so the automation survives the planned cloud migration with minimal rework. We also proposed a change and adoption stream, because efficiency gains only turn into a lower cost base when roles, process ownership and reporting are redesigned around the automation, and an optional managed service for running the platform. Delivered: - Proof of concept on live purchase orders with baseline metrics for ROI tracking - Production email-to-OrderWise order processing automation - AI Centre SKU normalisation model with continuous retraining - Action Centre review queue and customer clarification workflow - Secure OrderWise API integration with error handling - Orchestrator and Insights performance and exception dashboards - Enablement plan for internal IT ownership and a migration-ready integration design Results: - Proof of concept validated on live purchase orders and SKU data within 4 weeks, before any licence commitment - Production email-to-OrderWise automation live, removing manual order entry from the intake process - Customer SKU descriptions matched to internal product codes by an AI Centre model that retrains on exceptions - Ambiguous orders routed to Action Centre rather than guessed, with clarification emails to customers where needed - Peak-period volume handled without adding staff, with throughput and exceptions tracked in UiPath Insights - Internal IT enabled to own the platform, with the integration modularised for the planned OrderWise cloud migration ## Industries ### Financial Services https://ettgroup.ai/industries/financial-services ETT helps regulated firms cut the manual handling behind management reporting, reconciliation exceptions and evidence preparation. Every workflow is designed with source traceability, permissions and an approval route a reviewer can follow. ### Retail and eCommerce https://ettgroup.ai/industries/retail-and-ecommerce ETT connects order status, returns eligibility and service escalation to the systems that can actually resolve the request. Capacity is designed against your peak demand, and anything outside policy routes to a person. ### Manufacturing and Supply Chain https://ettgroup.ai/industries/manufacturing-and-supply-chain ETT helps manufacturing and supply chain teams reduce rekeying across purchase-order processing, supplier-document validation and exception routing. Extraction accuracy is measured, and anything the system cannot read with confidence goes to a person. ### Healthcare https://ettgroup.ai/industries/healthcare ETT helps healthcare teams reduce the administrative handling behind booking and rescheduling, incomplete referral information and administrative routing. This is administrative work, not clinical triage: decisions requiring professional judgement stay with clinicians. ### Logistics and Delivery https://ettgroup.ai/industries/logistics-and-delivery ETT connects delivery status, failed-delivery routing and customer updates to your operational systems. Repeat contact is measured rather than assumed, so you can see whether an answer actually ended the enquiry. ### Utilities, Oil & Gas https://ettgroup.ai/industries/utilities-oil-and-gas ETT helps utilities and energy operators reduce the manual coordination behind work-order preparation, inspection-document handling and outage communication. This is back-office and field coordination. Safety-critical authorisation and control of operational equipment remain with your own systems and people. ## Insights - [How to launch an AI voice agent in a regulated contact centre: a 10-step checklist](https://ettgroup.ai/insights/launch-ai-voice-agent-regulated-contact-centre): A 10-step checklist that runs governance alongside the build: choosing the call type, setting the baseline, disclosure, handover, identity and payments, testing, controlled release and the evidence you keep each month. - [The voice is the easy part](https://ettgroup.ai/insights/the-voice-is-the-easy-part): PolyAI says one UK bank's voice agent was built in about six weeks and governed in six months. Here is where the time goes, and what to put in front of your risk team in week 1. - [Containment is the wrong number](https://ettgroup.ai/insights/containment-is-the-wrong-number): In PolyAI case studies and NS&I's GOV.UK record, voice agents in regulated service lines contain 21% to 42% of calls. That is a good result. What matters more is what happens to the rest, and to the callers who ring back. - [Your voice agent speaks for you. Who signed off what it says?](https://ettgroup.ai/insights/who-signed-off-what-your-voice-agent-says): Disclosure, vulnerable callers, a route to a person, recording and payments: what the rules expect of an AI voice agent in a regulated firm, and a one-page sign-off sheet to complete before go-live. - [How to work out AI cost per task before you build](https://ettgroup.ai/insights/ai-cost-per-task): A step-by-step method for AI cost per task: model calls, tokens, hidden reasoning, price multipliers, checking and volume, with a worked example. - [EU AI Act Article 50 and 2 December: a checklist for UK and North American firms](https://ettgroup.ai/insights/eu-ai-act-article-50-checklist): What Article 50 of the EU AI Act asks of UK and North American firms, what changes on 2 December 2026, and a checklist to start now. - [An AI agent governance framework: a control template for regulated firms](https://ettgroup.ai/insights/ai-agent-governance-framework): A fill-in template for governing AI agents: register entry, autonomy tiers, approval gates, logging, identity, kill switch and staged rollout. - [How to measure what an AI summary leaves out](https://ettgroup.ai/insights/measure-ai-omissions): Accuracy checks catch what AI adds, not what it drops. A step-by-step method and table template to check an AI summary for missing information. - [Process mining, process intelligence or a discovery workshop: which comes first?](https://ettgroup.ai/insights/process-mining-vs-process-intelligence): Process mining, task mining, process intelligence and discovery workshops compared: what each tells you, what it needs, and which to start with. - [The AI evidence pack: ten records your regulator or auditor will ask for](https://ettgroup.ai/insights/ai-evidence-pack): What evidence do auditors and regulators need for AI? Ten records, who asks for each, what good looks like, and where firms usually fall short. - [Your 2027 AI budget will be wrong. Here's how to make it wrong by less.](https://ettgroup.ai/insights/your-2027-ai-budget-will-be-wrong): Gartner has published three different 2026 AI spend forecasts this year. Here's how to set a 2027 AI budget your finance team can defend. - [Brussels moved the deadline, not the risk](https://ettgroup.ai/insights/brussels-moved-the-deadline-not-the-risk): The EU delayed its high-risk AI rules to December 2027. The parts that didn't move already apply, and on 2 December 2026 more of them start to bite. - [Everyone counts what AI makes up. Almost nobody counts what it leaves out.](https://ettgroup.ai/insights/what-ai-leaves-out): In one study of clinical notes, AI left things out more often than it made things up. Why omissions need their own number, and how to count them. - [Field story: the logistics business that stopped building bots one at a time](https://ettgroup.ai/insights/stop-building-bots-one-at-a-time): More than 220 automations, each built from scratch. Here's what changed when the business agreed the parts before building any more machines. - [Seats or tokens: are your AI licences paying for themselves?](https://ettgroup.ai/insights/seats-or-tokens-are-your-ai-licences-paying-for-themselves): Most organisations now pay for AI twice: per seat for the assistants their people use, and per token for the systems their developers build. Nobody compares the two, so nobody knows which half is the expensive one. - [What actually drives an AI bill: the levers behind one number](https://ettgroup.ai/insights/what-actually-drives-an-ai-bill): An AI invoice arrives as a single figure, but it is the product of half a dozen separate decisions. Model choice, output length, hidden reasoning, caching and timing can each move the same workload's cost several times over. - [How to build an AI spend forecast your finance team will sign off](https://ettgroup.ai/insights/how-to-build-an-ai-spend-forecast): A step-by-step method for turning AI invoices into a forecast by workload: what to collect, how to measure each job, how to price the alternatives and how to present a budget that survives a finance review. - [What the AI hallucination cases mean for any firm that signs AI-drafted work](https://ettgroup.ai/insights/what-the-ai-hallucination-cases-mean-for-firms-that-sign-ai-work): Courts and tribunals on both sides of the Atlantic have now ruled on AI-generated errors. The pattern is consistent: the tool is never the defendant. Whoever put their name to the output answers for it. - [An accuracy score is not evidence: how to measure AI output so the result stands up](https://ettgroup.ai/insights/an-accuracy-score-is-not-evidence): "Our AI is 95% accurate" answers none of the questions an auditor will ask. Measured against what? Counting what? And what about the facts it left out? Here is what the research says a defensible measurement looks like. - [How to check an AI-written document against its sources, claim by claim](https://ettgroup.ai/insights/how-to-check-an-ai-written-document-against-its-sources): A step-by-step method for verifying a summary, report or letter drafted by AI: fixing what you check, breaking it into claims, finding the evidence, giving honest verdicts, looking for omissions and keeping a record that stands up later. - [What an AI proof of value should include before it reaches production](https://ettgroup.ai/insights/what-an-ai-proof-of-value-should-include): A proof of value earns its name by being able to fail. Here is what one has to contain — criteria set before the build, a real workflow, a defined failure path — and what it costs you when any of that is missing. - [What drives the cost of integrating voice AI with a contact centre](https://ettgroup.ai/insights/what-drives-the-cost-of-voice-ai-in-a-contact-centre): The licence is rarely the expensive part. Integration scope, intent count, language coverage and testing depth are what move the number — and each one is a decision you can make differently. - [How to calculate automation value without confusing time released with cash saved](https://ettgroup.ai/insights/time-released-is-not-cash-saved): Hours released and money saved are different claims, and presenting one as the other is the fastest way to lose a finance stakeholder. Here is how to calculate both, and how to say which one you have. - [When a rules-based workflow is enough, and when an AI agent adds value](https://ettgroup.ai/insights/when-rules-are-enough-and-when-an-ai-agent-adds-value): Not every workflow needs an agent. Deterministic automation is cheaper to build, cheaper to run and easier to assure — here is how to tell which one you are looking at, before you pay for the wrong one. - [How to evaluate AI retrieval quality and cost on your own documents](https://ettgroup.ai/insights/how-to-evaluate-ai-retrieval-quality-and-cost): Vendor benchmarks tell you how a system performed on someone else's corpus. Here is how to run the test on yours, which measures to keep apart, and the comparison most evaluations quietly avoid. - [The Cost of Asking: why AI bills rise as prices fall](https://ettgroup.ai/insights/the-cost-of-asking-why-ai-bills-rise-as-prices-fall): Token prices fell by roughly two thirds in a year while enterprise AI budgets grew about sixfold. The gap is volume, and the cause is structural: a language model has no index into a document, so every question pays to read everything. - [How to measure the ROI of enterprise AI](https://ettgroup.ai/insights/how-to-measure-roi-of-enterprise-ai): The ROI of enterprise AI is measured by operational outcomes — time saved, errors reduced, faster decisions and better service — not by activity or model performance alone. Defining success metrics before deployment is essential. - [What is conversational AI and voice automation, and where does it deliver value?](https://ettgroup.ai/insights/what-is-conversational-ai-and-voice-automation): Conversational AI and voice automation use natural language to handle interactions — answering questions, routing requests and triggering actions — while connecting to the systems behind them. The value comes when conversations lead to real operational outcomes. - [From pilot to production: how enterprises operationalise AI](https://ettgroup.ai/insights/how-to-move-ai-from-pilot-to-production): Moving AI from pilot to production depends less on the model and more on orchestration, data readiness and governance. A structured approach — orient, prove, govern, scale, compound — is what makes the difference. - [A practical guide to AI governance for enterprises](https://ettgroup.ai/insights/ai-governance-for-enterprises): AI governance is the set of policies, controls and oversight that keep AI systems safe, compliant and accountable as they scale. Done well, it enables AI adoption rather than blocking it. - [What is AI orchestration, and why does it matter for enterprise AI?](https://ettgroup.ai/insights/what-is-ai-orchestration): AI orchestration is the layer that connects AI models, data, systems, workflows and people so automation works reliably and under control. It is what turns isolated AI tools into operational systems. - [What makes business data AI-ready?](https://ettgroup.ai/insights/what-makes-business-data-ai-ready): An explanation of the data foundations needed for reliable automation, AI grounding and governed decision-making. - [Why AI projects stall at pilot stage](https://ettgroup.ai/insights/why-ai-projects-stall-at-pilot-stage): A look at why enterprise AI initiatives lose momentum and what needs to be in place before they can scale. - [What is agentic AI, and why should enterprises be paying attention?](https://ettgroup.ai/insights/what-is-agentic-ai): A practical introduction to agentic AI, orchestration and the move from AI experimentation to AI operation. ## Who runs ETT ### George McKenna, Chairman & CTO Eighteen years in enterprise IT before ETT, most recently leading solution sales at Trustmarque and running cloud sales and operations at Ultima Business Solutions. That was a career spent on the supplier side of large IT programmes, watching good ideas stall because nobody could say where to start, which is the problem ETT was set up to solve. He is accountable for ETT's technical direction. ### Ryan Carter, Group Chief Executive Officer Twenty years building and scaling technology ventures, most of it in secure communications and privacy technology. He has run global market development at Bittium, led corporate transformation as president of Vital Group of Companies, and founded two companies of his own before that. He leads the group, and is accountable for growth and commercial strategy across manufacturing, financial services, healthcare, retail and logistics. ### Richard Manser, Chief Delivery Officer Fifteen years delivering technology and automation, with a career rooted in quality engineering and a focus on making sure what gets built works in the real world. He has worked at the point where strategy meets delivery: designing solutions, building delivery capability and bringing the rigour of quality engineering to the way technology is built, governed and operated. At ETT, he owns how client work is scoped, delivered, evidenced and handed over, alongside its quality and commercial performance. His job is simple to describe: 'make sure ETT delivers what it promises'. ## Contact sales@ettgroup.ai. https://ettgroup.ai Headquarters: Dallas, Texas, United States. Other locations: London, United Kingdom; Vancouver, British Columbia, Canada. Map of the site: https://ettgroup.ai/llms.txt. Sitemap: https://ettgroup.ai/sitemap.xml