for investors and deal teams
Every pitch says AI now. The code tells the truth.
I'm Will Mitchell. I build AI systems for a living, and I can tell you from the inside: some 'AI-powered' companies are the real thing, and a lot of them are a thin wrapper and a demo that falls over. From the outside you cannot tell which one you're holding. From inside the codebase, you can.
Before you wire the money, I open the code, talk to whoever built it, and give you a straight answer: what's real, what's theater, and what it would take to fix.
01 / the offer
Technical AI Diligence
One deal or one portfolio company. Fixed fee, agreed before I start. Run it before you close, or in the first 100 days after.
For you
A straight verdict on every AI claim in the deal: build on it, kill it, or rescue it. What's actually making money, and what only works in the demo.
For the company CEO
A 100-day plan: stop this, fix that, build this first. Written to be executed, not filed.
I get repo access and time with whoever built the system. You get a written report where every finding names its evidence, so you can pull it back out at any board meeting and check what's still true. No hourly meter, no surprise scope.
02 / operator proof
I've sat on both sides of this table.
Bought and operated
I was co-CEO of a PE-backed rollup. We started with a $1.5M platform acquisition and grew the company to roughly $54M in enterprise value. I know what a 100-day plan looks like when you actually have to live it.
Built in production
As a fractional AI co-founder I built a HIPAA-aware healthcare SaaS from first commit to production: the architecture, the agents, the evals, the compliance. Shipped software in a regulated industry, not a slide deck about one.
I do this every day
I run StartupBros, PRBot, and House of Vibe, and I build the AI systems inside them myself. My own market-research system keeps 130 stocks under continuous AI coverage, with every claim checked against sources. Reading AI systems is not a service I bolted on. It is what I do all day.
03 / methodology
What I actually look at.
No black box. This is the checklist, in plain words. Hold my report to it.
- What they claim
Every AI capability in the deck and the data room, written down as something we can actually check.
- Does it touch money
Which of those claims show up in revenue or costs, and which only show up in demos.
- What's in the code
The actual repositories, read directly: how it is built, what it depends on, and what breaks when it grows.
- Do they measure it
Whether anyone can prove the AI works, or whether it just feels like it works.
- Can the data carry it
Whether the data layer and the day-to-day workflows can support the roadmap being sold.
- What happens when things change
Models get deprecated, vendors reprice, APIs break. Who would notice, and what it costs when they do.
You do not get a deck. You get a report where every finding names its source and can be re-checked after I am gone.
What one finding looks like (illustrative example)
- Claim
- "The recommendation engine drives a third of revenue," per the pitch materials.
- Evidence
- Order attribution over the trailing 90 days: 4.1% of completed checkouts touched a recommendation surface.
- Falsified by
- Re-run the attribution query at close and at day 100; the claim fails if attributed revenue stays under 10%.
- Verdict
- Kill the claim, rescue the pipeline. The revenue claim is theater. The event pipeline underneath it is sound and feeds two items in the 100-day plan.
04 / expansion path
Where this goes next
Some investors keep me around after the deal as an ongoing AI partner across their portfolio. That only opens up after we have done a diligence engagement together. By then you are not taking my word for anything. You have seen the work.
05 / start
Have a deal on your desk?
This page is for investors: deal teams, operating partners, independent sponsors, or anyone putting real money behind an AI-heavy company. Building your own company instead? Head to the main site