AI readiness audit: find the work worth automating first
- One week, fixed scope
- Ranked use cases with the reasoning written down
- A data and risk review of the top candidates
- One pilot recommendation with a cost range
Why start with an audit
Most AI projects that disappoint do so for ordinary reasons: the data the feature depends on is incomplete or out of date, nobody owns the result, the cost of a wrong answer was never discussed, or the tool was chosen before the problem. An audit puts those questions first, while changing course is still cheap.
It is also the honest answer to being pitched AI products. After a week you know which of your processes would benefit, what it would take, and what you should not do yet. Our article on what AI can actually do for a small or mid-size business describes the five kinds of job where AI tends to pay off.
What you get in one week
Ranked use cases
The candidate uses we found, ordered by value, effort and risk, with the reasoning written down so your team can challenge it.Data and risk review
For the top candidates: where the data lives, whether it is accurate and allowed to be used, what a wrong answer would cost, and which privacy rules apply.One pilot recommendation
The pilot we would start with, what success looks like in numbers you already track, and what we would deliberately leave out of it.A cost range
A build cost range and the monthly running costs that drive it (model usage, hosting, maintenance), or a recommendation to buy an existing product instead.How the audit works
- Scoping call. Before the week starts we agree the goals, the processes in scope, the constraints and the people we should talk to.
- Interviews. Short conversations with the people who do the work (support, sales, operations, finance) to find tasks that are repetitive, rule-heavy or depend on looking things up.
- Systems inventory. What you run (store platform, CRM, helpdesk, documents, spreadsheets) and how information moves between those systems today.
- Data sampling. With your permission and under a non-disclosure agreement, we look at real samples to see whether the information an AI feature would rely on is complete, current and usable.
- Read-out. A working session where we walk through the ranking, the risks and the recommended pilot, followed by the written report.
Where a question can only be answered by trying something, the report says so and makes it part of the pilot.
Questions we work through
For every candidate use case, the audit answers the same set of questions before anything is recommended.
- Can the data be reached, and is it accurate, current and allowed to be used?
- How many cases a month, and how much staff time do they take today?
- What does a wrong answer or a wrong action cost, and who would notice?
- What happens when the AI cannot handle a case: who picks it up?
- Which number will show that it works, and what is it today?
- Which systems would it read from or write to?
- Which privacy rules and customer promises apply?
- Who owns the result after launch, and how would you switch it off?
Who this is for
- Businesses with a process that eats staff hours: support questions, order exceptions, document intake, reporting.
- Teams that tried a chatbot or AI tool and want to know why it disappointed.
- Owners being pitched AI products who want an independent view before signing.
Who it is not for
- Research projects or training a model from scratch.
- Organisations where nobody can own the result after launch.
- Adding AI to the roadmap for its own sake: if the answer is "not yet", the report says so.
What it costs and what happens after
The audit is a one-week engagement with an agreed scope. We quote it after the scoping call, because the number of processes and systems in scope changes the effort.
After the read-out you can take the report to any vendor, run the pilot with your own team, or ask us to build it. We build grounded AI assistants, agent workflows with approval steps, AI features for online stores and AI search visibility work.
What we have built
We have built a retrieval-grounded (RAG) assistant, an MCP server that lets AI applications use a business system through defined tools, embeddings-based semantic search, and an AI workflow agent whose actions a person approves. We describe them in general terms here, without client names or numbers; ask us for the details on a call.
The rest of our background is the systems AI has to plug into: we have built websites, online stores and mobile apps since 2005, today mostly on Laravel, Node.js, Magento, Shopify, WooCommerce and AWS, and in October 2026 we moved our own site from WordPress to Astro on AWS.
Further reading from our team:
What we will not do
The same rules apply to every AI project we take on. They are the reason some projects end with a smaller scope, and a few with a recommendation to wait.
- Put an agent into production that acts without a person who can approve, stop and undo what it does.
- Train or fine-tune a model on your data, or send it to a provider that does, without your written consent.
- Publish AI-generated text on your site or send it to your customers without a person reviewing it first.
Frequently asked questions
How long does the AI readiness audit take?
One week of work after a short scoping call. The week covers interviews with the people who run the processes in scope, an inventory of your systems, a look at samples of the data, and a read-out session with a written report.
What do we need to provide?
Two or three hours with the people who do the work in scope, access to samples of the relevant data under a non-disclosure agreement, and one person who can take decisions on the recommendations.
Do we have to use Impinge for the pilot?
No. The report is yours and is written so that your own team or another vendor can act on it. Some audits end with a recommendation to buy an existing product, or to wait.
Will our data be used to train AI models?
No. Data you share stays under the non-disclosure agreement, we only use approved business tools with client data, and we never train or fine-tune a model on your data without your written consent.
How much does the audit cost?
We quote it after the scoping call, because the number of processes and systems in scope changes the effort. You know the price before the week starts.
Start with one week, not a roadmap
Tell us which processes take the most time and we will scope the audit with you.
