← Common Problems
The Problem

“The target says it's an AI company — we can't tell if that's real”

"Is this an AI company or a company with an AI slide?" is now a diligence question in its own right. The pitch deck won't answer it; the architecture, the data, and the team will. Here's the diagnosis.

An AI slide is not an AI moat

In 2026, nearly every target has an AI narrative. The diligence question is whether the AI is a genuine differentiator or a thin wrapper on someone else's model that any competitor could assemble in a weekend. The deck rarely distinguishes the two, and a deal team that can read a business but not a codebase can't tell from the outside whether the data assets are defensible, whether the architecture can support the roadmap management is selling, or whether the "AI" is doing anything a few API calls couldn't.

Getting this wrong is expensive in both directions: you can overpay for an API wrapper dressed as a platform, or pass on a target whose data moat is real but undersold. The signal lives in places a deck doesn't show — proprietary data, model differentiation, evaluation discipline, deployment maturity, and key-person risk.

Assess AI substance as a first-class diligence dimension

The fix is to treat AI readiness and differentiation as a named, first-class part of technical due diligence: independently assess whether the target's AI is defensible, whether its data is a genuine moat, and whether the architecture can actually support the roadmap — then translate that into investment-grade findings the IC can act on, with risks and likely remediation cost quantified.

What good looks like

An independent read on whether the AI is a real differentiator or a thin wrapper
The data moat assessed for genuine defensibility, not asserted
Architecture checked against the roadmap management is actually selling
Findings packaged for the investment committee, not just the CTO
The service that resolves this

Technology Due Diligence
Questions

Common questions

How do you tell a real AI company from an API wrapper?

You look where the deck doesn't: whether there's proprietary data the model is trained or grounded on, whether the model is genuinely differentiated or a thin layer over a third-party API, whether the team has evaluation discipline, and whether the architecture can support the roadmap. A defensible data moat and real eval rigor distinguish an AI company from a company with an AI slide.

Can our deal team assess this internally?

Deal teams can assess the business, but judging whether AI is a defensible differentiator requires reading the code, the data, and the architecture with senior engineering experience — and an independent read that an investment committee can trust over an internal one on a deal everyone is motivated to close.

Other problems we solve

“Our reports don't reconcile across the firm”

“We rolled up 25–30 companies and the data lives in as many systems”

“An analyst spends days every month rebuilding the same report”

Sound familiar?

Tell us what you're seeing and we'll tell you straight whether — and how — we'd fix it.

Schedule a Call