Front door
AI Production Audit
A focused technical read of what stands between an AI pilot, feature, agentic workflow or production system and a reliable production outcome. If the answer is that you do not need us, the report will say so.
No charge. 30-min call. Written report. No sales script.
What the call covers
Six questions about the system you actually have. Not every one will apply, and the ones that do will not get equal time: the call follows the problem you bring.
- What working has to mean
- The failure modes that actually matter to your business, and the behaviour the system has to hold before it can ship.
- Evaluation
- Whether evaluation exists, what it measures, and whether it is connected to anything that can stop a release.
- Release controls
- What currently blocks an unsafe change, at which point in the pipeline, and on what evidence.
- Observability and drift
- What you can see about production behaviour after deploy, and how you would know the system had moved.
- Recovery
- What happens when the system is wrong: fallbacks, rollback, and who is paged.
- Ownership
- Who operates this after launch, and whether the operating model matches the team you actually have.
How it works
- 01
Book the call
A 30 minute slot. Three questions sent in advance so the call starts with context rather than discovery.
- 02
The call
A structured technical conversation about the specific system, workflow or pilot you need to get to production. Led by a senior engineer close to the work.
- 03
Written report
Within five working days. Where the work sits against production, what stands between it and shipping, and what the path looks like, ranked.
- 04
Whatever you decide
Proceed, pause, or do nothing. The report is yours either way, and it is written so your team can act on it without us.
What this is not
The audit is a way in, not a gate. It does not commit you to a programme, a retainer or ongoing operation, it does not run all seven delivery stages, and it does not produce the full set of production evidence. It gives you a focused read on the production problem you bring, not a full engineering programme.
Questions
What does it cost?
Nothing. There is no charge for the call or the report.
Do we have to commit to anything afterwards?
No. The report is yours regardless of what you decide next, and it is written to be useful to your own team on its own.
What if the answer is that we do not need you?
Then the report says so. That is a more useful outcome than a proposal you did not need, and it is the reason the report is worth reading.
Who is on the call from your side?
A senior engineer. The audit is not delegated, and it is not a qualification call run by someone who could not answer a technical question about your system.
Does this commit us to a large programme?
No. Engagements are scoped to the problem, and most begin with a single one. Ongoing operation is an option rather than a default.
Our AI runs on Azure OpenAI, Anthropic, Bedrock, or self-hosted. Does that matter?
The provider changes implementation details, but not the core production questions. We still need to understand how the system behaves, how it is evaluated, what can stop an unsafe release, what happens when it fails, and who owns it after launch.
Will you sign an NDA?
Yes. A standard mutual NDA is available on request.
How is this different from a sales call?
The call produces a written technical report, not just a proposal or next-step conversation. You can use that report internally even if we never work together.
Find out what stands between your AI work and a reliable production outcome.
No charge. 30-min call. Written report. No sales script.