The Assessment
The AI Production-Readiness Assessment
A fixed-fee engagement that tells you, and your board, whether the AI inside your institution is safe to run, and leaves you with the file an examiner asks for. A clear answer and a prioritized path to fix what is not, in two to three weeks.
The Offer
Fixed scope. Fixed price. A board-ready answer.
No open-ended discovery and no six-figure transformation. One engagement, one deliverable your leadership can act on.
- Scope
- Fixed
- Investment
- $15K to $25K
- Timeline
- 8 to 10 consulting days over 2 to 3 weeks
- Delivery
- Remote-first
What you get
- An assessment against the five GRADE pillars. Governed, reliable, safe agent behavior, run like production, and tested, each graded against how a mid-market institution actually operates.
- The board-level GRADE Scorecard. A single production-readiness deliverable your board, auditors, and exec team can read without a translator.
- A prioritized remediation path. The gaps that matter, ranked by production and exam risk, with a concrete sequence for closing them.
- The examiner file. The nine documents examiners and internal auditors ask for first, from the AI inventory and vendor due-diligence file to the kill-switch test record and issue register, in the format your institution already uses for IT exams.
How It Works
Four steps, start to readout.
Concrete, low-burden, and time-boxed. Your team stays focused on the product while I do the assessment.
Scoping call
A working session to agree on the one or two AI systems in scope, the questions your board is asking, and what a passing grade needs to mean for your firm.
Evidence gathering and system review
I review the architecture, prompts, guardrails, logs, and eval coverage, and talk with the people who build and operate it. The burden on your team is a handful of interviews and read access, not a project.
Scorecard and findings
I grade the system against the five GRADE pillars, document the gaps in plain language, and rank them by production and exam risk rather than by how interesting they are to engineers.
Readout with your board or exec team
A live walkthrough of the GRADE Scorecard and the remediation path, in language a non-technical board can act on. You leave knowing what is safe to ship, what is not, and what to fix first.
Who It Is For
Mid-market financial services heading to production.
Banking, insurance, and fintech firms with an AI pilot, an internal copilot, or a vendor AI feature that is moving from experiment to something customers and regulators will touch. If it is about to carry real work, it is in scope. That includes the feature your core vendor switched on and the chatbot someone in operations is already using.
You do not need an ML platform team or a compliance department. You need to know, with evidence, that the system will not break, leak data, or hand the examiner a finding once it is live.
It is built for a regulated context. The Fed's April 2026 model-risk guidance puts generative and agentic AI outside its scope and says you still have to govern them; the assessment gives you the framework no one handed you. Vendor management, IT risk, and fair lending are treated as practical considerations, not as a reason to panic.
After the Assessment
A fractional CTO retainer, when you want it ongoing.
Most firms find the Assessment surfaces more than a weekend of work. The natural continuation is a fractional CTO retainer: ongoing senior capacity to work through the remediation path, guide the next build, and keep the system examiner-ready as it grows. It runs in the $5K to $15K per month range depending on scope, and it graduates directly from the Assessment, so there is no second discovery to pay for.
A two-to-five-day readiness snapshot on a single system is available when the scope is still uncertain, so you can start narrow and expand once the picture is clear.
Find out where your AI actually stands.
Start with a scoping call. We agree on what is in scope and what a passing grade means for your institution before any work begins.