Enterprise organizations whose AI procurement runs eight to thirty-six months. A two-week, fixed-fee audit doesn't fit that approval process, and forcing it wastes both our time.
Compiling the next screen…
Compiling the next screen…
A two-week audit of your AI system. What's broken, what it's costing you, and what to fix first. $6,500, fixed.
Thirty minutes. Not refundable, and credited against the audit if you move forward.
They come from teams whose demo works and whose production system does something nobody can account for. If more than one line is yours, the model is not what is broken.
Five deliverables. The findings report is the product; the rest supports it.
| Deliverable | What it is |
|---|---|
| Findings report | What's broken, ranked by what it costs you, in plain language |
| Evaluation gap analysis | What you'd need to measure to know the system actually works |
| Cost and latency trace | Where the spend and the slowness come from |
| Remediation proposal | Fixed-fee scope for the work the findings justify |
| Readout call | One hour, your team in the room, questions answered live |
Access to the repository, the logs, and the model dashboards. Ninety minutes each with three or four people on your team. And two meetings on your calendar before day one: the findings readout and the remediation walkthrough.
That last part carries the two-week promise. The window holds only when the meetings are scheduled going in. Without them, this becomes a longer, slower engagement, and the price and timeline move to match.
Before day one
Scope signed. Access to the repository, the logs, and the model dashboards. Both review meetings already on your calendar.
Weeks one and two
One person reads the whole system and holds all of it at once. Ninety minutes each with three or four of your people.
End of week two
Findings report, evaluation gap analysis, cost and latency trace, remediation proposal. Then the two meetings you booked: the findings readout and the remediation walkthrough.
$6,500, fixed. Scoped before you sign, so the fee can't drift once we start.
Half on signature, half on delivery of the findings report. The $150 diagnostic call is credited against the first half.
The call fee is not refundable. The credit is good for six months from the day you pay it, and it applies to the audit or to any larger engagement that comes out of it. Pay, talk, then decide. If six months go by without you moving forward, the credit lapses.
In 2009, at Ingram Micro, I built the Roadmap Quickstart Process: a method business and technical groups used to produce grounded, business-driven implementation roadmaps. I also adapted the TOGAF Architecture Development Method to Ingram Micro's own software development methodology, with its own deliverables, process documentation, and engagement guidance.
This audit is that process, aimed at an AI system instead of an enterprise roadmap. It was built for and used by a Fortune 100 distributor's own business and technical groups, not assembled for this offer.
Four facts, all public and checkable without asking me.
[David] consistently delivered high-quality code, demonstrating meticulous attention to detail and a commitment to writing clean, efficient, and scalable software.
[David's] knowledge related to security, communication of issues and solutions, availability and attitude were wonderful.
Two kinds of company shouldn't book this call.
Enterprise organizations whose AI procurement runs eight to thirty-six months. A two-week, fixed-fee audit doesn't fit that approval process, and forcing it wastes both our time.
Seed-stage founders who built the pilot themselves. At that stage, the fix happens at your own desk at midnight, not through a consultant.
An AI evals hire takes three to six months to find and ramp. The audit tells you what to hire for first, so the headcount decision isn't a guess. It doesn't compete with the hire. It makes the hire cheaper to get right.
No. That badge certifies vendor product knowledge, not diagnostic independence. I'm not paid more when you use more tokens, so I have no reason to tell you to use more of them.
True, and for a two-week assessment, that's the point. One person reads the whole system and holds all of it at once, instead of losing context across a handoff between specialists. Scale is a remediation-phase problem, and the remediation proposal can name subcontractors then.
David Bleeker
Principal Consultant • Applied AI Engineering
Thirty minutes, $150. Not refundable, and credited against the audit fee if you move forward within six months.