An AI audit that ends with working code
We look at the AI decisions in front of you, tell you plainly what’s worth building, build one piece of it, and stay on call for six months. Free for founders in the accelerator programs we support.
Not in one of those programs? Our standard AI opportunity audit is $5–10k fixed, with a full money-back guarantee.
Who it’s for
- Pre-seed to Series A teams deciding where AI belongs in the product.
- AI-first teams whose product works in demos but not reliably with real users.
- Founders without a senior AI engineer yet who want a second opinion before they hire or build.
What you get
- 01
Scoping call
You bring 1–3 concrete decisions. We audit only those.
- 02
Decision memo, within 2 weeks
For each decision: a do / don’t / buy verdict, the recommended architecture, the main risks (accuracy, cost, latency, data and privacy) and an effort estimate. If you want us to build any of it, a fixed price is included. No obligation.
- 03
One thing built
We ship one working artifact into your repo — usually an evaluation suite for your LLM feature (50–100 test cases with automated scoring you can run in CI), or a prototype of the top recommendation. It stays yours.
- 04
Walkthrough call
With your founders and engineers.
- 05
Six months of office hours
One hour a month with a senior engineer to review architecture, models, evals or AI hiring as you grow.
Questions we usually answer
- Should we ship this LLM feature now, or wait?
- RAG, fine-tuning, or neither?
- Build, buy, or skip?
- Why does our agent work in demos and fail with real users?
- Which models should we use — and what will they cost at 10× our usage?
- Where can AI cut our operating costs, and where will it only add risk?
What this isn’t
- 01
Not a generic report
We only answer the decisions you bring.
- 02
Not a slide deck
You get a written verdict and code that runs.
- 03
Not a sales funnel
The memo stands on its own, whether or not you ever hire us.
Evgeny Smirnov
PhD in astronomy, 20+ years in software, active researcher who publishes on LLM evaluation in peer-reviewed journals. He works on client projects himself. The habit he brings from research is simple: measure before you claim.
You work directly with Evgeny and senior engineers with 15–20 years of experience. No account managers, no juniors.
Track record
- products shipped
- 100+
- raised by our clients
- $500M+
- average client relationship
- 5+ yrs
- building since
- 2008
- AdvisorEngine
- trivago
- And 100+ more
Timing and terms
- We usually start within 3 business days of your request. If many requests arrive at once, we schedule them in the order received.
- One audit per company.
- Your time: two calls and a few async answers.
- NDA on request. We need read-only access only for the part we build.
Is it really free?
Yes, for founders in the accelerator programs we support. We do it because the best way to start a long working relationship is to be useful first.
What do you need from us?
A scoping call, a few async answers, a walkthrough call, and read-only access to the code we build on.
Who owns what you build?
You do. The memo and the code are yours.
What happens after the audit?
Six months of monthly office hours. If you want us to build something from the memo, the fixed price is already there. If not, that’s the end of it.
We don’t have AI in the product yet. Is this still useful?
Yes. Deciding where AI doesn’t belong is half of the audit.