AI Audit & Due Diligence
We tell you whether a target's technology and AI can really deliver, before you invest. Run by engineers who build production AI, not analysts.
Trusted by
What we assess
Six areas, and we go deepest on the one most often oversold: the AI.
AI and data
Is the AI real, or a thin wrapper around a model someone else trained? We check what the models actually do, where the data comes from and whether the rights to use it are clean, how defensible the AI is, and whether the roadmap can be built.
Architecture and scalability
Will the platform hold as the business grows, or does scale mean a rewrite? We map the architecture, the real bottlenecks, and the cost of keeping it running.
Code and engineering practices
We read the actual code, not just the diagrams: quality, test coverage, security hygiene, and how cleanly the team ships and reviews changes.
Security and compliance
How customer data is handled, the security posture, and the gaps (GDPR, SOC 2, data residency) that can stall a deal or surface after close.
Team and delivery
Who actually builds the product, whether the team can deliver the plan you are underwriting, and where the key-person risk sits.
Product and roadmap
How mature the product really is, how it holds up against competitors, and what it will actually take to hit the plan.
Our process
A focused process built around your investment thesis.
Scope
We align on your thesis and the questions that actually matter, then lock the scope and data-room access.
Deep review
We work through the data room, the code, and the architecture, run the AI and security checks, and sit down with the CTO and key engineers.
Report and readout
You get a written report and a live readout: an executive summary, a scorecard, risks ranked by severity, and a value-creation view.
Most audits run from 3 to 4 days for a focused read to 2 to 3 weeks for a full review, depending on scope and deadline.
Deliverables
A report your committee can act on, not a slide deck.
Executive summary
A one-page read for the investment committee, focused on what actually affects the decision.
Scorecard
Every area rated from solid to needs-work, so the picture is clear at a glance.
Risk register
Risks ranked by severity, each with its business impact and what it takes to fix.
AI assessment
A straight read on the AI: what is real, what is a wrapper, and how defensible it is.
Value-creation plan
The AI and engineering moves worth making after the deal closes.
Our edge
The audit is run by the people who build this for a living.
Built by builders
Your audit is run by engineers who ship AI in production, not analysts who have only read about it.
AI-native depth
Multi-agent systems, RAG, LLM evaluation, and the data layer underneath. We know exactly where modern AI breaks.
Fast and senior
Founder-led, from 3 to 4 days to 2 to 3 weeks depending on scope, with straight answers your committee can act on.
Case study
A private equity firm was evaluating a fintech ahead of an investment and needed to judge the technology, the AI strategy, and the team inside a short diligence window. We reviewed the data room, the architecture, and the code, pressure-tested the AI and data-layer thesis along with its own internal AI usage, and ran two deep-dive sessions with the CTO. The firm got a clear read on the real technical moat, the risks to watch, and where AI could extend the product, in time for the decision.
Questions from funds
How long does an audit take?
From 3 to 4 days for a focused read to 2 to 3 weeks for a full audit, depending on scope and deadline.
What access do you need?
A data room, read access to the codebase, and time with the CTO and a few engineers.
What is in the report?
An executive summary for the committee, a scorecard, a risk register ranked by severity, a straight read on the AI, and a value-creation view.
Can you assess early-stage or pre-revenue AI startups?
Yes. With early-stage companies we focus on whether the team and the AI can deliver the thesis, not just on what is already shipped.
Do you only look at AI?
No. We cover the whole stack: architecture, code, security, team, and product. The AI is just where we go deepest.
Can you help after the deal?
Yes. We can scope the value-creation work and build the AI systems we recommend.
Let's talk about your project
Build custom AI systems for your requirements
Need AI built for your exact requirements? Let's discuss your project.
Deployed for startups, scale-ups, national research centers and Big 4 consulting firms