See what one workflow needs before AI agents take it on
- Nine questions about one workflow, in about three minutes.
- Every reason in the result links to the page on this site it comes from.
- Data is the one hard prerequisite. Scattered data and consequential decisions shape the design.
Free · about three minutes · result on this page and by email
A first step, and the reasons behind it
The result is short enough to read in a minute and to forward to the person who needs to agree.
- 01
The first step that fits
Named and priced exactly as it is published, with how long it takes. Where something has to be settled before any step is bought, the result names that first and the step that follows it. - 02
What is ready
The answers that work in your favour, each with one sentence on why it matters and a link to the page it comes from. - 03
What to settle first
The answers worth sorting out before or alongside the work, named one by one, each with the same kind of link. - 04
Design notes
What your sector and the kind of decision involved mean for the design. They never change the step or its price. - 05
What the step gives you
The published deliverables of the step, and how it ends.
Want to hear how this reads for your own workflow? Ask James Rooney on a 30-minute call and he will go through it with you.
Talk it throughHow the assessment reaches its answer
Each answer raises one named reason, and fixed rules turn the reasons into a step. Every reason is a sentence one of our pages already publishes.
Each answer raises one named reason
Every reason is one sentence taken from a guide, a service page or a sector page, with a link to that page. Prices and deliverables come from the same data the pricing and service pages read.
The rules run in a fixed order
A workflow that could affect safety in an industrial setting comes first, because the industrial and energy page says to start somewhere else. Several candidates point to the AI Readiness Audit. A need for a working method, or an outcome with no pound value as the only gap, points to Learn. A working pilot with an owner and people who can take part points to the 90-day outcome programme. Everything else points to the Agentic Proof of Concept.
Five things come before any step
Data that does not yet exist in a usable form, a process nobody owns, work that is not written down, people who cannot be freed, and experts who disagree about what correct is when a wrong result is consequential. The result names the first of them and the step that fits once it is settled.
Nothing is scored
A maturity model tells you where you sit against other organisations, which is interesting and rarely actionable. So the result names what is ready and what to settle, and no answer produces a verdict against building. Whether a workflow is worth automating at all turns on volumes and the cost of an error, which only your operation knows.
If one of these rules does not fit your workflow, ask James Rooney on a 30-minute call how he would read your answers.
Talk it throughWhat each question looks for, and the page behind it
Each question maps to something a Tenhaw page publishes. Follow the link for the full reasoning.
| Question | What it looks for | Page behind it |
|---|---|---|
| What decision are you trying to make? | Whether you have one workflow in mind, several candidates, a pilot that already works, or a need to learn a working method first. This answer chooses between the four published first steps. | Who a proof of concept is right for |
| Where does the information this workflow needs live today? | Whether the data is reachable, scattered, held on-premise or not captured yet. Data is the one hard prerequisite, so only data that does not yet exist in a usable form has to be settled first. | What a proof of concept needs |
| Is the work written down, and can the expert take part? | Two of Tenhaw's three conditions for a proof of concept: requirements in writing, even if they are years out of date, and an expert who can be in the room on days four and five. | The three conditions |
| How quickly can a qualified person tell whether a result is right? | Whether a result can be verified inside a day, and whether your experts agree on what a correct answer is. | Document intelligence guide |
| What happens when a result is wrong? | What the exception path costs. There is no universal accuracy number, and quoting one is a warning sign. | Document intelligence guide |
| What would the system be allowed to do? | Whether it reads, proposes, makes reversible changes or takes irreversible actions, which decides the permissions and confirmations to design. | Tool calling and system integration guide |
| Who owns the process and its result? | Whether one named person can agree its value and what counts as working, and whether anyone has put a pound value on the outcome. | Business case guide |
| Can your engineers and the people who do the work take part? | Whether your engineers can pair with ours and the people who do the work can test, with time protected during build and adoption. | Who the 90-day programme is right for |
| Which setting is this in? | Which sector notes to add. It never changes the step or its price. | Sectors |
Not sure which of your workflows to answer these for? Bring the two you are weighing to a 30-minute call and talk them through with James Rooney.
Talk it throughBring the result, or bring the question.
Thirty minutes with James Rooney, free. If you have run the assessment, your answers arrive with the booking and the call starts from them. If you have not, describe the workflow and talk through which first step fits.
Free · 30 minutes · James confirms call requests personally.
No time that suits? Email hello@tenhaw.com. Calendar not loading? Open it on cal.com.
Describe the work
Describe the work in a paragraph. James replies with a rough scope and budget range within two working days.
Questions about the assessment
Is the AI readiness assessment the same as the AI Readiness Audit?
No. The assessment is free: nine questions about one workflow, with the result on the page. The AI Readiness Audit is a whole-organisation assessment at £44,000 fixed over four weeks: it explores four areas, tests two or three candidates and gives your board a costed plan.
The assessment points to the audit when you have several candidate workflows and need to choose between them.
Does the readiness assessment give our workflow a score?
No. It returns the first step that fits and two lists of named reasons, what is ready and what to settle first, each linked to the page it comes from. There is no percentage, rating or traffic light.
The reason is the one we give about maturity models: they tell you where you sit against other organisations, which is interesting and rarely actionable.
Can a free readiness check tell us whether a workflow is worth automating?
No. Whether a workflow is worth automating at all turns on volumes and the cost of an error, which only your operation knows.
What the assessment can show is what is ready, what to settle first and which published step fits. A proof of concept ends with an evidence-based go/no-go decision, and the AI Readiness Audit tests two or three candidates as working prototypes against real data in your tenancy.
What does the readiness assessment suggest if our data has not been captured yet?
It names getting the data into a usable form as the thing to settle before buying anything, then shows the step that fits once that is done. Data is the one hard prerequisite, and where it does not yet exist in any usable form, a proof of concept is the wrong place to start.
Data that exists but is scattered or held on-premise does not hold the step back. It is listed as something to settle, with a note on what it changes, such as where a proof of concept sits in its £20,000 to £55,000 band.
Will the readiness assessment rule out a workflow that affects customers or money?
No. A wrong result that reaches a customer, money or a regulator is listed as something to settle and adds design notes, and on its own it does not change the first step. Agents take decisions that are high-volume, observable and cheaply reversible; humans retain decisions that are consequential, contested or hard to undo.
If your experts also disagree about what a correct result is, agreeing that comes first, because the ground-truth set is built with the people who own the decision, before anyone argues about models. In an industrial setting, a workflow that could affect safety is the one case where the assessment suggests starting with a different workflow.
Why does the readiness assessment ask what the AI system would be allowed to do?
Because what a system may do decides the permissions and confirmations it needs. Tenhaw labels every tool by consequence and reversibility before it is exposed: read, reversible write, and irreversible or consequential action.
A system that changes things on its own needs write tools that are single-purpose with narrow typed arguments. An irreversible action either requires a human confirmation that names what is about to happen, or is raised as a request for a person to execute.
What does the readiness assessment suggest when a pilot already works?
The 90-day outcome programme, when one named owner can agree the outcome's value and your engineers and users can take part. It is £120,000 for two practitioners or £180,000 for three, fixed, paid in three monthly instalments of £40,000 or £60,000.
If nobody owns the whole process, or people cannot be freed, the assessment names that as the thing to settle first. If nobody has put a pound value on the outcome, it points to Learn. Without engineers to pair with ours, it points to a proof of concept first and lists the engineers as something to settle.
Why does the readiness assessment ask how quickly a result can be checked?
Because a result nobody can verify inside a day cannot be user-tested in weeks three and four of a proof of concept. Quick checking is the test we ask you to add to our own three conditions: data that already exists in usable form, requirements in writing even if they are years out of date, and an expert who can be in the room on days four and five.
If your experts disagree about what a correct result is, the assessment lists building a ground-truth set with the people who own the decision. Done properly, that is a fortnight of work rather than a phase.