Our work

What changed for the teams we worked with.

A selection of the work we do. Clients are described by sector rather than named, because most of this work sits inside regulated businesses and the detail belongs to them. Each engagement below is real, scoped, won and delivered. Where we publish outcome figures, they have been measured with the client, not projected.

Case studies

Delivered engagements

Logistics and supply chain

Turning 220 siloed automations into a reusable framework

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.

New automation delivery time cut by 60 to 70 percent

Read the case study
Social care and public sector technology

An AI assistant for social workers, kept entirely inside the platform

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.

Live in-platform AI assistant with zero external data processing

Read the case study
Financial services

Evidence before automation for a financial services operator

Discovery-led engagement mapping how priority processes actually run across systems and people, standing up UiPath Automation Cloud, and producing a ranked roadmap to act on it.

Read the case study
Healthcare and legal services

Email-to-case-system automation as a group-wide foundation

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.

£120k+ annual saving, 5 to 8 FTE of manual effort removed

Read the case study
Financial services

Taking the year-end squeeze out of finance audit and reconciliation

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.

Manual audit effort cut by up to 80 percent, payback in around 7.5 months

Read the case study
Manufacturing and distribution

From inbox to ERP without the re-keying

Automating purchase order intake from customer emails and PDFs into OrderWise, with machine learning to resolve inconsistent product codes and an Action Center queue for anything ambiguous.

Manual order entry removed from the intake process

Read the case study
Sample deliverables

The two documents that decide whether an automation works

One defines what gets built and what happens when it cannot complete. The other defines how you will know whether it made a difference. Both are agreed with you before anything is built.

Sample

Workflow blueprint

The document a workflow is designed from. Nothing gets built until every row here is agreed, because each blank is a decision someone would otherwise make silently during delivery.

Scope

Workflow
Inbound order status enquiry, web chat and email channels
Volume
Measured over four weeks before design, split by channel and hour
Out of scope
Refunds, order amendments, anything requiring a credit decision

Systems and permissions

Reads
Order management, warehouse despatch, carrier tracking
Writes
Case record only. No write access to the order itself
Identity
Dedicated service account, scoped to the reads above, rotated on your schedule
Logging
Every read and write, with the identity, source and timestamp

Decision points

Automated
Confirm despatch status, confirm expected delivery date, resend tracking
Requires a person
Anything outside the returns window, any goodwill gesture, any complaint
Approval route
Named service team queue, with the case history attached

Failure behavior

Source unavailable
Tell the customer what is known and when the next update is due. Escalate. Do not estimate
Order not found
Hand to a person with the reference the customer gave
Tracking stale
State the last confirmed scan and its time rather than inferring a location
Sample

Measurement plan

Agreed before the build starts, not chosen afterwards to suit the result. A baseline taken after go-live is not a baseline.

Baseline

Period
Four consecutive weeks before any change, including one peak week
Source
Contact platform records and order management, not a manual estimate
Excluded
A known outage week, excluded by agreement and stated in the report

Measures

Resolution rate
Request completed, not merely answered. Reported for automated and assisted paths separately
Repeat contact
Same customer, same order, within seven days
Handling time
First contact to resolution
Cost per resolved contact
Total cost to serve divided by contacts resolved

Comparison

Period
Four weeks after a two-week stabilisation window, matched for seasonality
Confounders
Promotional activity, carrier disruption and volume shifts recorded and reported alongside the result
What we will not claim
Released time as cash saved, unless a headcount or cost line actually changed

These are purpose-built samples for an order status workflow. They are not redacted client work and they are not a client result. The rows in your blueprint would carry your systems, your policy and your thresholds.

What we will publish

The standard a result has to meet before it appears here.

We would rather have an empty results page than a persuasive one built out of proposals and pilots that never reached production.

  • The client is named with written approval, or described specifically enough to be useful and generally enough to stay anonymous.
  • ETT's role is stated exactly, including which parts a delivery partner delivered. We do not publish a partner's work as ours.
  • The stage the engagement actually reached is stated. An engagement still in delivery publishes what it is designed to deliver, not a result.
  • Baseline and result use the same measure over comparable periods, with the calculation shown.
  • Modeled results are labelled as modeled. Projected production savings are never attached to work that did not reach production.
  • The source system, measurement period, sample size and limitations are held on file for every figure we publish.
Explore by area

Where this kind of work usually starts

Workflow automation

Connecting systems so information stops being carried by hand, with exceptions routed to a person.

View solution

Conversational AI

Resolving routine customer requests and handing complex ones over with the context attached.

View solution

Strategy and readiness

Deciding which opportunity is worth building first, and what it would take.

View solution

Data readiness

Making approved information usable for a specific decision or workflow.

View solution

Discovery and classification

Establishing what data exists, how sensitive it is, and which sources AI may use.

View solution

Security and governance

Named security ownership for an AI deployment, with the controls and decisions written down.

View solution

Have a similar problem? Talk to us.

Tell us the process you would start with. We will talk through what the blueprint and the measurement plan would look like for it.

Request an AI session How an engagement works

Region & currency

Changes spelling, terminology, the data-protection regime named in our notices, and the currency used in indicative figures. ETT is based in London — this is not a local office or a price in your currency.