Acuity AI Advisory

Examples of our work · Reviewed October 2026

What does working with Acuity look like?

Sometimes the issue is time lost to repetitive work. Sometimes it is an AI draft that still needs too much fixing, or a board that needs clearer oversight. These examples show the questions we worked on and the material we prepared to help clients address them.

The examples are anonymised and summarised from our work. Client records and commissioned training materials remain private. Each example distinguishes the work delivered from any outcome that would need to be measured afterwards. Anonymised accounts of Claude implementation work, including which Claude product was used, are on the AI at Work case studies page.

Workflow assessment

Deciding which manual work to automate

A professional team was repeating administrative checks around its existing systems. Replacing those systems was a separate decision, so the immediate question was which improvements could stand on their own.

Acuity examined the process and produced a prioritised recommendation. It separated work suitable for the existing Microsoft environment from changes that depended on a future system. It also identified steps where the risk of a failed automation outweighed the time it might save.

What the work produced

A workflow assessment and implementation priorities, with feasibility conditions made explicit. Estimates of potential time savings were planning inputs; they were not measurements of time already recovered.

Discuss a repetitive workflow →

Copilot adoption

Making long-document drafts easier to check

A team could already get Copilot to produce a draft, but longer source material led to omissions and more editing. A fluent answer was not enough for the work it needed to support.

Acuity prepared practical guidance for structuring the input and checking the result against the underlying records. The work included a comparison method so the team could assess the draft against an existing human-reviewed output.

What the work produced

A tailored working guide and a defined evaluation approach. The target was a more useful first draft; the guide itself was not evidence that a particular time saving had been achieved.

Talk about disappointing Copilot results →

Board advisory

Connecting AI oversight to decisions the board makes

A leadership team needed to consider how AI affected the information it relied on and the decisions for which it remained accountable.

Acuity prepared a board session around the organisation's operating context. It covered the questions directors should ask about evidence, ownership and the review of AI-assisted analysis, with attention to proportionate oversight.

What the work produced

A tailored session deck and decision prompts for board discussion. This demonstrates the advisory work prepared, without implying that a session alone made the organisation compliant.

Discuss AI oversight with us →

AI policy advice

Checking that policy fits the tools people use

An organisation needed its written AI policy to reflect the tools it was authorising, while taking account of confidentiality and supplier terms.

Acuity reviewed the policy alongside the proposed use of the tools and prepared a position note. The advice distinguished the location of processing from contractual safeguards and identified wording that needed to be reconciled with authorised use.

What the work produced

A position note and proposed policy changes for the organisation's own review. Legal and data-protection decisions remained with the responsible functions.

Discuss a policy that needs updating →

What changes when the manual steps go?

Two examples of the work. These simplified walkthroughs explain the process using illustrative records, not client data or promised savings.

A client reminder cycle

The handler checks a spreadsheet, sends reminders and removes clients who have replied. The same checks happen again at the next stage.

With the workflow: the handler reviews the due list, sends the batch and sees confirmations reflected in the next list.

Discuss a process like this

Example: review before sending

Client A · documents outstanding
Include in reminder batch
Client B · receipt confirmed
Exclude from reminder batch
Client C · handler follow-up
Hold for review

The person approves the action. The workflow handles the repeatable steps.

A legal operations check

A fee earner works through records individually, checks the required information and chases missing items before the next stage.

With the workflow: complete records and exceptions are separated, so the person can concentrate on the decisions and checks that need judgement.

Read the engagement stories

Example: an exception list

Record A · required fields present
Ready for handler review
Record B · missing reference
Return for completion
Record C · conflicting figures
Check against source

A clear exception is more useful than an answer that hides uncertainty.

Is something similar getting in your way?

Tell us what you are trying to improve and which systems you already use. We can discuss where advice, training or a workflow change would help, and agree the scope before work begins.

Talk to us