Who this audit is for
Founders and heads of engineering with a product already in users' hands, looking at AI as the next feature surface but unsure of the shape, the cost, or whether it ships in this quarter or next.
- Existing SaaS or web product. Real users, real data, real load. AI is a feature you are adding, not a product you are inventing.
- A specific use case in mind. Search over your corpus, an in-app assistant, classification or extraction at scale, an agent that completes a workflow.
- Internal AI experiments that have not become production features yet, and a board or leadership team asking why.
- Prior LLM prototypes or vendor demos that need an honest second read before a budget gets approved.
What you walk away with
A 6 to 8 page written brief, delivered as PDF and as a private live page. Single page executive summary for the board and the CFO. Technical appendix for the engineering team.
- Feasibility and shape of the AI feature in your specific product context. What it looks like at the system boundary, where the model lives, what the failure modes are.
- Production cost model. Per call economics, projected monthly burn at expected volume, where the line items shift if traffic grows tenfold.
- Vendor recommendation. Model, retrieval strategy, evaluation framework. Reasoning, not a matrix. One named option per layer with the trade-off explained.
- Named risks and what kills each one. Compliance flags called out where they apply. Regulated industry multipliers if the use case sits in HR, healthcare, finance, or legal.
- Go or no go recommendation with three explicit options on the table. Ship it as scoped, re-scope and ship a smaller version, or do not ship and what to do instead.
A 60 minute walk-through call with the engineering team is included. The brief belongs to you. If you take it to a different shop to build, that is a clean outcome, not a failure.
What is not in this audit
- No code written. No prototype. No proof of concept demo. The audit decides whether to build, not what to build with.
- No vendor procurement on your behalf. The audit recommends. You sign the contracts.
- No fine-tuning or training data work. That sits inside the integration build, not the audit.
- No ongoing operations after the brief is delivered. The integration build absorbs that, or the advisory retainer covers it long-term.
The two-week shape
- Kickoff call with the engineering team and the product owner.
- Read the existing stack, the data layout, the user volume.
- Map the AI feature surface against where it lives in the product.
- Mid-week checkpoint. If the audit is not going to land useful, we walk and refund.
- Build the production cost model. Per call, monthly burn, growth scenarios.
- Choose the vendor stack. Model, retrieval, evaluation framework.
- List the risks. Surface compliance flags where they apply.
- Deliver the brief. 60 minute walk-through with the engineering team.
After the audit
If the brief says ship, the integration build picks up where the audit ends.
The integration build is the engagement that puts the AI feature into production. Six to ten weeks, fixed scope, 20 percent milestone holdback at the acceptance gate. From €25K, capped at €50K. Above that ceiling we re-scope or refer out.
If the build signs within 45 days of audit handoff, the €10K audit fee is credited toward the build invoice. The credit is named in the SOW, not held back as a discount lever.
Frequently asked
Why pay for an audit when I can book a free discovery call?
The discovery call exists. 30 minutes, no charge, useful for deciding whether this is even the right shop for you. The audit is the next step, where money changes hands so the analysis goes deep enough to land in front of your board. A discovery call cannot read your stack and cost the build. The audit can.
What if the audit recommends not to build?
That is a clean outcome. The board now has a written reason, with numbers, for parking the AI feature this quarter. Most no-go recommendations come paired with a smaller, less expensive thing worth doing first. The cost of the audit is recovered the moment a wrong-shaped build is avoided.
Is this an AI readiness assessment?
It is, with a specific shape. AI readiness assessments often stop at maturity scoring. The audit goes further. It costs the build, names the vendor, and produces a recommendation specific enough to act on. Readiness without a recommendation tends to gather dust.
What happens at the kickoff if the audit is not going to land?
If the kickoff call surfaces that the use case is too early, too vague, or sits outside what this audit can usefully assess, we say so. The fee is refunded and we part on good terms. The clause exists because mismatched audits help neither side.
Who delivers the audit?
Sagar Sutaria, directly. Studio engagement, no account manager layer. Currently CTO at Relokate (Berlin) alongside the studio, with seven production platforms shipped across HR, healthcare, legal, wellness, and consumer in the last 18 months.
Book the audit call
30 minute call. Describe the use case, the product, what is on the line. Decide together whether the audit is the right next step or a different shape of work fits better.
Book a 30 minute call →