The AI Due Diligence Checklist
10 key topics AI where can help your Due Diligence
I’ve been talking to a lot of angels and syndicate leads recently about how they actually use AI in their investment process.
One conversation in particular stuck with me.
I spoke with Pjotr van Schothorst (who runs PVS Investments). He’s a full-time professional angel, deep tech focused, technical background, with decades of investment experience. Pjotr evaluates deals for a living and doesn’t have a team of analysts behind him. Just a $20/month AI subscription (you can guess which) and a systematic approach.
I had Pjotr list off how he uses the model, what topics he covers and what questions he asks. That’s already interesting and it resonated, but the more I listened, the more I realized what solo syndicate leads need is a repeatable process that turns AI from a novelty into an actual edge. More on that in future posts.
Here’s the checklist, with context on why each step matters.
1. Bust the “we’re the only one” myth
2. Build a competitor comparison table
3. Fact-check every claim
4. Go deep on the technology
5. Run a ‘techno-economic’ analysis
6. Check the market dynamics
7. Reality check the patents and their relevance
8. Scan the legal documents
9. Investigate the team
10. Find and contact domain experts
Let’s dive in.
1. Bust the “we’re the only one” myth
Every pitch deck has a version of it. “We’re the only company doing X.” “No one else has solved this.” “We have no direct competitors.”
Copy-paste that one-liner into any AI. Ask: who else does something similar? You’ll almost always get a list back. Sometimes it’s other startups. Sometimes it’s established companies, incumbents who’ve been doing a version of this for years.
This isn’t about catching founders in a lie. Most genuinely believe they’re unique. But a competitive landscape that takes an analyst days to compile now takes thirty seconds, and it gives you much better questions to ask in the next call.
2. Build a competitor comparison table
Once you have a list of comparable companies, ask AI to organize them. Founded when? How much raised? Revenue if known? Growth trajectory? What’s their specific approach?
This structured view is something pitch decks never give you. You go from “trust me, we’re special” to an actual map of the space. It changes the quality of your questions entirely, and it takes about two minutes.
3. Fact-check every claim, one by one
Every statement in a pitch deck that’s presented as fact, check it. That should be a core discipline.
“We’re the only cobalt processor in this region.” Are they? “The market for X is $50 billion.” Is it? “Our technology is 10x more efficient.” By what measure, compared to what?
AI won’t catch everything, but it catches a lot, and fast. The pitch deck is a marketing document, your job is to figure out where the marketing ends and the reality begins.
4. Go deep on the technology
For deep tech, climate, or hard science startups, you need to understand whether the core technology can actually be commercialized. This is where AI has gotten surprisingly good.
Ask it to explain the underlying process. How does this technology work at a basic level? What are the known challenges in scaling it? What’s the current state of the art? Who else is working on similar approaches?
You’re not trying to become a techie or scientist overnight. You’re trying to understand enough to ask smart questions and catch obvious gaps between what’s claimed and what’s real.
5. Run a ‘techno-economic’ analysis
This was the step that surprised me most. AI doesn’t just summarize text anymore, it runs calculations (and you can trace the assumptions).
Feed in the startup’s claims about unit economics, production costs, and market pricing. Ask AI to model whether the numbers work; what’s the market price for this product; how does it develop over time; what are the underlying assumptions, and are they realistic?
This is a big time saver for climate and deep tech deals where the science and the economics are tightly coupled. It won’t replace a proper financial model, but it will tell you whether the basic math holds up before you spend hours building one.
6. Check the market dynamics
Related to the above, but worth its own step. Ask AI about the specific market the startup is targeting. What are current prices for the key inputs or outputs? How have those prices evolved over the last decade? What drives price volatility?
A startup claiming they’ll profitably mine rare earth metals at a specific price point? AI can pull historical pricing, identify the major producing countries (often more than the pitch deck suggests), and flag whether the assumptions are optimistic, realistic, or fantasy.
7. Reality check the patents and their relevance
“We hold 12 patents” sounds impressive, but which ones matter?
Feed the full patent documents into AI. Ask for a plain-language summary: what does this patent actually protect? Is it relevant to what the company is currently building, or is it a legacy from an earlier pivot? Does the chief science officer’s research history actually align with the technology?
I’ve heard of cases where founders’ PhDs were in completely different fields from the company they started. The patent portfolio looked impressive on paper but had zero relevance to the current product. AI spots this in minutes because it can cross-reference the patent claims against the company’s stated technology.
Some startups list patents they’ve moved beyond. They keep them in the deck because having patents looks good, but the actual IP has evolved. That’s not necessarily a red flag, it’s just something you want to know.
8. Scan the legal documents
Term sheets, shareholder agreements, side letters. These used to require expensive lawyers just for a first read. AI changes the economics completely.
Feed the documents into AI and ask: Is this a standard term sheet? Is this a standard shareholder agreement? Flag any unusual terms. What deviates from the norm? What are the risk factors?
Most terms in most deals are boilerplate. What you care about are the outliers, the clauses that an experienced lawyer would circle in red. AI is remarkably good at spotting those, and it gives you a focused list to discuss with your actual lawyer instead of paying them to read forty pages of standard text.
9. Investigate the team
This is the one area where AI is least useful, at least for now. People assessment is still ‘manual’. But AI still helps around the edges.
Start with the basics: are all the people listed in the pitch deck actually listed as working at the company on LinkedIn? You’d be surprised how often they’re not. “Oh, he’s an advisor.” “Oh, she’ll join full-time once we close the round.” These are red flags worth knowing about before you invest, not after.
For science-heavy startups, check the team’s publications. If your chief science officer’s entire research history is in a different field from the company’s core technology, that’s a risk factor. AI can quickly surface publication histories and tell you whether someone’s academic background actually matches what the company is building.
And use public registries. In the UK, Companies House lets you look up any director’s track record for free. How many companies have they founded? How many are still active? A pattern of dissolved entities doesn’t necessarily mean the person is bad, but it does mean the “successful serial founder” narrative in the pitch deck needs questioning.
10. Find and contact domain experts
One thing AI is good at: helping you find the right people to talk to. Once you understand the technology, ask AI to help you identify experts in the field, people who’ve published on the topic, worked in the industry, or have relevant experience.
LinkedIn works well for this. Look for people with a track record on the specific subject, not the big-name VCs or famous professors (who often have the least useful perspective), but the working professionals who actually know the space from the inside.
An expert who’s genuinely excited about a startup’s approach is a strong signal. An expert who says “this is interesting but I’ve never seen it work at scale” is equally valuable information. Either way, you want outside perspective beyond the pitch deck.
Making it systematic
The real insight here isn’t any single step. It’s the systematic application of all of them to every deal.
Most angels use AI opportunistically: they check something when it feels off, or they ask a question when they’re curious. Pattern recognition is a key human skill. The investors I’ve spoken to who get the most value from AI use it as a checklist.
This is low cost and not time consuming. It’s a fraction of what it would cost to hire analysts, lawyers, and accountants for even a preliminary screening.
This can be a shift for solo syndicate leads and small operations. You’re not replacing judgment, you still need experience, pattern recognition, and gut instinct to make the final call. What you’re replacing is the grunt work of information gathering and ‘reality checking’: the hours spent reading patent filings, cross-referencing team bios, and manually checking market claims.
---
*This is part of a series on practical tools and workflows for syndicate leads and angels. If you’re building or running a syndicate, check out SyndicateOS.ai.*