What is an AI readiness audit?
An AI readiness audit tells you where AI will pay back in your business and whether you are ready to use it there. Most assessments answer that from a few interviews with managers. Ours counts the work itself. The Evolve Workflow Audit is how we do it.
A fixed-fee engagement that answers one question properly: where does your team’s week go, and what is it costing you. Every member of staff in scope documents their own repeating work, so the picture is a census rather than a sample. You end up with two things. The first is full documentation of every process in the business, department by department, which most companies have never had in one place and which keeps its value whether or not you automate anything. The second is the same work turned into annual hours and cost, ranked by what is worth automating first. You can buy it on its own, and it also runs at the front of every AI automation, agentic AI, and bespoke software project we deliver.
Most failed AI projects fail in the first two weeks of choices
Picking the wrong workflow, missing a compliance dependency, or underestimating the change effort is far more expensive than any technical mistake that comes later. By the time the build team is involved, the difficult decisions have already been made.
The Workflow Audit is the safeguard. We do the observation work that is almost always skipped, sitting with your people, watching the routines, mapping the actual flow of work, before any model is selected, any code is written, or any platform decision is made.
The output is the document the engagement is then planned and priced from. It is also the artefact that, on its own, often pays for itself: clients describe the experience as the only consultancy meeting where they left with a clearer picture of their own business than when they walked in.
The four phases
The Workflow Audit covers four observable phases. The underlying tooling that drives them is proprietary and remains so. What clients see is the output of each phase.
Listen
Everyone in scope: not a sample of the team, and not just the managers describing the work. Each person documents the routines that consume their week, the workarounds they have built, and the moments they describe as “the bit that always slows us down”, with the time and frequency attached. Fifteen to thirty minutes each, in their own words, whenever suits them.
Map
We build an explicit picture of each priority workflow, how it moves across people, systems, and decisions. Hand-offs, exceptions, decision gates, and the data each step produces or consumes. Drawn from observation, not from a process diagram somebody wrote three years ago.
Score
Every candidate opportunity is evaluated against a consistent rubric. Business impact, technical feasibility, regulatory risk, change-management cost, time to value. The same lens applied to every option, so prioritisation is defensible rather than political.
Sequence
A phased roadmap that makes the order of work obvious. Quick wins to build confidence and free capacity, foundational investments in data and platform where needed, strategic plays scheduled once the foundations exist. The sequencing reflects your appetite for change.
What you walk away with
Concrete, board-ready deliverables. Every output is designed so your leadership team can make informed decisions about where to invest.
Every process, department by department
A complete written record of what your business does all week, grouped by department and searchable across all of them. Most companies have never had this in one place. It is worth keeping whether or not you automate a single thing.
Ranked opportunity register
Every process captured with its annual hours and cost, then ranked. Where several people described the same job the total is corrected, and both figures are shown so the adjustment is visible rather than assumed.
The money, stated three ways
What the repeating work costs a year, how much of that is addressable once feasibility is taken into account, and a realistically recoverable range. Plus what the whole thing amounts to in full-time equivalents.
Tiered by how much person it needs
Every process sorted into work that can run with nobody involved, work where AI does it and a person checks, and work that genuinely needs a person throughout. Hours and addressable cost against each.
Where to start, not just what is big
Every process placed on a pain against feasibility map, and a recommended sequence with dependencies. The biggest number is rarely the right first move, and the map is what shows you why.
Systems and software register
Which platforms consume the time across the business, what the subscriptions cost a year, and how much of that spend sits against work that could be automated. Drawn from the tools your people named.
Board-ready, and auditable
A summary a non-technical leadership team can decide from, the whole dataset as a PDF and a spreadsheet, and a stated data-quality position: who did not respond and what the total therefore excludes. Steps touching personal data are flagged at capture and redacted by default.
The data itself, not a locked dashboard
Everything exports as a spreadsheet whenever you want it. Take it to your own IT team, to your software supplier, to another agency, or to us. Nothing is held back to make the next phase conditional on using Evolve, and the register is scoped and priced from the same document whoever does the work.
See a real one
Two figures from the sample report. Northgate Logistics is a fictional company, so nothing is redacted: 18 people across four departments, 14,747 hours a year of repeating work costing £357,158, and the reasoning behind every number.
A handful of processes carry most of the cost
Six of 21 processes account for 72% of the annual cost. That is what makes the sequencing obvious.
View as a table
| Process | Hours a year | Cost a year |
|---|---|---|
| Chasing delivery paperwork | 2,500 h | £60,000 |
| Daily route planning | 2,250 h | £54,000 |
| Driver debrief and defects | 2,200 h | £52,800 |
| Answering "where is it?" | 1,688 h | £38,813 |
| Damage and shortage claims | 1,125 h | £25,875 |
| Booking timed deliveries | 1,050 h | £24,150 |
Almost none of it truly needs a person throughout
Only 1% of the hours genuinely need a person start to finish. Most of the rest is work a person would check rather than do.
- Runs without a person16%
- AI does it, a person checks83%
- Needs a person throughout1%
View as a table
| Kind of work | Processes | Hours a year | Share |
|---|---|---|---|
| Runs without a person | 6 | 2,315 h | 16% |
| AI does it, a person checks | 14 | 12,313 h | 83% |
| Needs a person throughout | 1 | 120 h | 1% |
The full report runs to nineteen pages: every process costed, a pain against feasibility map, the software register, and a recommended sequence. Tell us where to send it.
The five pillars of AI readiness
AI readiness comes down to five things: the work, the data, the systems, the people and the governance. The audit reports on each, so you know what has to change before AI goes in, and what can go in now.
1. Work
Which tasks repeat, how often, and what they cost. This is the pillar most assessments guess at, and the one the census measures directly.
2. Data
Whether the documents, records and messages the work depends on are digital, findable and clean enough for a model to read.
3. Systems
Where the work lives today, and whether those systems can be connected to, or need replacing first.
4. People
Who does the work, who will check the AI, and what training they need so adoption sticks after launch.
5. Governance
An AI policy, a register of where AI is used, and the controls UK GDPR and, where it applies, the FCA, SRA or ICO expect before anything touches client data.
For regulated firms that need to satisfy a board or a regulator, the governance pillar goes deeper: a written policy, a model register and the evidence trail. See AI governance and policy.
It also runs at the front of every build
Bought on its own, the audit is a decision document. Bought ahead of a build, it becomes the brief. Either way we will not pick the technology before we know where the time goes.
AI Automation
Every AI automation engagement opens with an audit, so the thing we automate is the thing that costs you money.
Learn moreAgentic AI
For agentic work the audit is non-negotiable. Pointing an autonomous system at the wrong workflow costs far more than looking first.
Learn moreBespoke software builds
The audit becomes the build brief, so the first phase is scoped from what your people do rather than a wish list.
Learn moreFrequently asked questions
What is the Evolve Workflow Audit?
A fixed-fee audit of the work that repeats. Every member of staff in scope documents their own routines, so you get a census rather than a sample. We turn that into annual hours and cost per process, correct for the same job being described by several people, and rank what is worth automating first.
How long does a Workflow Audit take?
About three weeks from launch to report. Each person spends 15 to 30 minutes documenting their own routines, at a time that suits them, and can come back to it. Nobody has to sit in a workshop or fill in a spreadsheet.
Is the Workflow Audit an AI readiness assessment?
It is our AI readiness assessment, built on evidence rather than interviews. It covers all five pillars: the work, the data, the systems, the people and the governance. Regulated firms that need a board paper get a deeper governance section on top.
What do we get at the end of the Workflow Audit?
Full documentation of every process in the business, department by department, with the time each takes and the systems it touches. On top of that, a prioritised opportunity register, recommended sequencing with realistic timelines, and compliance and governance notes for any opportunity touching client data or regulated decisions. Yours to keep, whatever you decide to do next.
Do we own the data from the Workflow Audit, and can we use it without Evolve?
Yes to both. The register and the underlying data are yours, and export as a spreadsheet whenever you want. Take them to your own IT team, your software supplier, another agency, or us. Nothing is held back to make a second phase conditional on using Evolve, and anyone can scope and price the work from the same document. If the audit says the return is not there, we will tell you that and you will still have documentation worth having.
Will the audit disrupt our day-to-day operations?
No. Nobody has to be taken off the floor for a workshop. Each person documents their routines in their own time, in 15 to 30 minutes, and the analysis happens on our side once the responses are in.
How do you avoid counting the same work twice?
Deliberately and visibly. When several people describe the same process, adding their hours together would inflate the finding, sometimes by a factor of three. We detect the overlap, adjust for it, and show you both the raw total and the corrected one so you can see exactly what was taken out and why. A number you cannot interrogate is not worth having.
Is the underlying methodology proprietary?
Yes. The tooling and frameworks that drive the audit are proprietary to Evolve. What clients see is the output: a sharp, defensible, and immediately actionable view of where to invest.
Start with a Workflow Audit
Most engagements begin here. Book an initial alignment call and we will scope a Workflow Audit shaped to the size of your business and the departments you most want to understand.