AI Workflow Reliance Review

You know what your workflow does. Could you defend it if challenged?

Knowing what your workflow does isn't the same as having the evidence to defend it.

The AI Workflow Reliance Review is a focused, asynchronous review of AI-supported workflows. Submit your workflow and receive a one-page AI Reliance Map showing where AI influences action, who relies on the output, what evidence exists, and what would be difficult to defend.

Submit a workflow for review See what the review includes

Asynchronous review

Each workflow is reviewed separately.

Output

One-page AI Reliance Map.

Data approach

No sensitive records, client files, patient data, or internal documents required.

First review

No meeting required, although clarification may be requested if needed.

What the review does

It maps where AI-supported reliance happens.

The AI Workflow Reliance Review maps where AI enters a workflow, what the output influences, who relies on it, what action follows, what review happens, and what evidence exists.

It is designed for workflows where AI may influence:

a decision
an action
a referral
an escalation
a communication
a record or note
a prioritisation
a case outcome

Each workflow is reviewed separately and returned as a one-page AI Reliance Map.

What you receive

A concise review output you can actually use.

AI Reliance Map

One-page AI Reliance Map per workflow.

Reliance view

Clear view of where AI reliance happens.

Visible evidence

Evidence currently visible from the workflow description.

Evidence gaps

Gaps that may make the workflow harder to reconstruct.

Defensibility concerns

Areas that may be difficult to explain if challenged.

Practical suggestions

Focused improvement suggestions, with optional clarification before completion if needed.

How it works

Describe the workflow, not sensitive source material.

You do not need to upload policies, internal documents, client records, patient data, employee data or confidential case material.

To start the review, you complete a short workflow submission form describing:

where AI is used
what the AI output does
who relies on it
what action follows
what review or sign-off happens
what evidence is currently retained

The review is designed to work from a practical workflow description, not from sensitive source material.

Workflow descriptions are treated as confidential and used only for the review.

If anything is unclear, I may contact you for clarification before completing the review. This helps avoid assumptions and keeps the final AI Reliance Map useful.

Each workflow is reviewed separately and returned as a one-page AI Reliance Map.

What not to include

The review does not need sensitive records.

Please do not submit personal data, client records, patient information, employee records, commercially sensitive documents, credentials, system exports, prompts containing confidential material, or live case details.

The review does not need them.

A clear description of the workflow is usually enough to identify where AI reliance happens, what evidence exists, and what may be difficult to defend if challenged.

This review is intentionally low-data. You are asked to describe the workflow, not upload sensitive records. If further context is needed, I will ask for clarification rather than guessing.

Workflow descriptions are treated as confidential and used only for the review.

Example output

A one-page view, not a long audit report.

The AI Reliance Map is not a long audit report. It is a one-page view of where AI reliance happens and what may be difficult to defend.

Example extract from a fictional workflow

Workflow reviewed: AI-assisted customer complaint triage

AI use: Incoming complaints are summarised and given a suggested urgency rating before being reviewed by a team member.

Action influenced: Priority queue placement and escalation decision.

Review area	What the review found
Where AI enters	AI summarises the complaint and suggests an urgency rating before human review.
Reliance point	The team member uses the AI summary and urgency rating to decide whether the complaint is escalated.
Human review	A human reviews the output, but the workflow does not show what they are expected to check before relying on it.
Evidence currently available	Final priority category is recorded. Original complaint is retained.
Evidence gap	The record does not show whether the human reviewer agreed with, changed, or challenged the AI urgency rating.
Defensibility concern	If a complaint was wrongly deprioritised, it may be difficult to show whether the issue came from the AI output, the human review, or the escalation criteria.
Suggested improvement	Add a light-touch review field: "AI urgency accepted / changed / rejected" with reason required only when changed or rejected.

In this example, the organisation can show the final priority decision, but not how the AI output influenced that decision. The workflow has human review, but the review is not evidenced.

Pricing

Pricing is per workflow.

Single workflow review: £450.

Volume pricing: £350 per workflow for five or more workflows submitted together.

Each workflow is reviewed separately. This keeps the review focused, proportionate and useful.

More detailed operational reviews can be scoped where a workflow needs trigger assessment, oversight review, evidence gap analysis, remediation actions or an executive summary.

Submit a workflow for review

Submit a workflow for review.

This review is intentionally low-data. Describe the workflow, not sensitive records. If further context is needed, I will ask for clarification rather than guessing.

Each workflow is reviewed separately. You can submit one workflow to start, or multiple workflows if you already know the areas you want reviewed.