Institutional investors have always had an edge in private markets. They’ve got teams of analysts, access to expensive data tools, and polished processes that solo investors or lean family offices could never match.

But that’s changing thanks to AI.

Anyone can now run a high-quality investment diligence process, the kind that once required a $120K p.a. analyst, with a smart prompt and a few plug-and-play tools.

That's better insights and faster decisions at much lower costs.

This guide shows you how to build that system yourself.

This is what due diligence looks like

Institutional due diligence is deep, but slow and bloated. One investment memo requires:

  • 2 analysts working 40+ hours
  • 12+ data sources (PitchBook, Crunchbase, LinkedIn, etc.)
  • Dozens of pages of notes, screenshots, and formatting
  • Another 2 weeks to finish the investment committee deck

That’s fine if you’re running a $500M fund with full-time staff. But if you’re investing your own money, working with a small team, or managing a <$30M fund, it just doesn’t scale.

You either waste time doing it all yourself, or skip steps and rely on your gut. Neither is ideal.

How does AI-powered diligence work

Instead of doing all the grunt work, you design a system that does it for you.

You don't replace your judgment, but remove the slow, repetitive parts so you can focus on decisions.

With AI tools like ChatGPT or Claude, you can now:

  • Pull market comps
  • Analyze funding history
  • Summarize team bios
  • Flag risks in legal docs
  • Access financial statements
  • Track hiring slowdowns or PR red flags
  • Create a clean, investor-ready summary

And it all takes less than 15 minutes.

The 3-part stack for fast, clear, AI-driven diligence

Here’s how to set it up.

1. The prompt engine

This is your system’s brain. A strong prompt can handle half an analyst’s job. Some of the key modules include:

  • Company overview (from website or Crunchbase)
  • Team background (bios, LinkedIn)
  • Funding timeline and valuation
  • News + PR sentiment
  • Market and competitor snapshot
  • Risk flags (churn, lawsuits, negative press)

You can combine these into one big prompt or break them into stages.

2. The data layer

AI is only as smart as the data it sees. Publickly available data is a great place to start, but for more sophisticated outcomes, you’ll need to plug in sources like:

  • Crunchbase or PitchBook APIs
  • Web scrapers
  • PDF tools for investor decks and financials
  • LinkedIn bio extractors (ethically and rate-limited)
  • Google Alerts or RSS feeds for live news

This setup gives you the same depth as institutional teams at a fraction of the cost.

3. The output layer

Here’s what you get once the AI does its job:

  • Clear, bullet-style executive summary
  • Visual market maps and funding timelines (via plug-ins)
  • Custom follow-up questions for your diligence calls
  • Highlighted risks with source links

Instead of 40 hours of research, you’ve got a clean 2-page briefing, ready in under an hour.

What this looks like in real life

Here’s how I use this system today:

I’ve built a plug-and-play prompt stack + tool system that runs everything above.

All you need to do is copy and paste the prompt below into your AI model (ChatGPT or Claude)

Prep work: User Thinking models only. Replace the {COMPANY_NAME} with the company name you want to run the research for. Add more context if needed by replacing {OPTIONAL_CONTEXT}.

PROMPT:

You are a senior private markets investment analyst preparing a family office grade due diligence report on a private company using only publicly available information and any browsing tools you have. Be skeptical, verify claims, and do not guess. Clearly label: Fact vs Company Claim vs Third Party Claim vs Analyst Inference. Prioritise primary sources (registries, filings, regulators, audited accounts, patents, court records, official press releases). Cite sources for all material statements with access date. If key data is missing or paywalled, state it and propose free alternatives or verification steps.

INPUT
Company name: {COMPANY_NAME}
Optional context (stage, geography, sector, ticket size, thesis): {CONTEXT}

DELIVERABLE
Produce an IC ready due diligence report for family offices/UHNWI, with clear headings, tables, and a final recommendation with conditions and next steps.

PROCESS (follow in order)

  1. Entity resolution
  • Confirm correct legal entity, jurisdiction, incorporation identifiers, HQ, domains, subsidiaries, brands. Note name collisions and how resolved.
  1. Source capture
  • List the key sources checked (company site, registries, regulators, filings, reputable media, IP databases, litigation/enforcement, social channels).
  1. Company and product
  • What it does, who it serves, pricing/packaging cues, product walkthrough from public info, proof points (customers, case studies, certifications).
  1. Market and traction
  • Category definition, ICP, use cases, demand proxies (traffic, reviews, app data, hiring), growth signals, retention or churn signals if available.
  1. Competitive landscape
  • Identify direct competitors, substitutes, and “build vs buy”.
  • Include a competitor table: competitor, segment, differentiators, pricing model, strengths/weaknesses.
  1. Business model and economics
  • Revenue streams, distribution channels, unit economics signals (CAC, payback, GM, churn) using filings or transparent proxies with ranges.
  • Summarise financials from filings if available (revenue, margins, cash, debt, liabilities). Flag auditor qualifications, late filings, liens/charges.
  1. Team, ownership, governance
  • Founders/executives background checks, credibility, departures, board/investors if disclosed, ownership/cap table signals, related party risks.
  1. Legal, regulatory, and risk
  • Licences/permissions, compliance posture (AML/KYC, GDPR, PCI, SOC2 etc if relevant), litigation, adverse media, reputational issues.
  • Provide a risk register table: risk, category, likelihood, impact, mitigants, diligence actions.
  1. Valuation and exit (high level)
  • Closest public comps and logic, any known private rounds, valuation range (if feasible) with assumptions. Identify plausible acquirers and exit routes.
  1. Diligence action plan
  • Tailored data room request list, management interview questions, customer reference plan, and top gating items.

OUTPUT FORMAT
A) Executive summary (bullets): opportunity, key positives, key risks, open questions.
B) Main report sections aligned to steps 3–9.
C) Tables: timeline, competitor table, ownership/funding table (if possible), risk register.
D) “Red flags and contradictions” (blunt).
E) Recommendation: Invest / Watch / Pass, with conditions and next steps.

Start now by resolving the entity for {COMPANY_NAME}, then proceed through the sections.

What this system can’t (and shouldn’t) do

It won’t:

  • Decide on the quality of the founder
  • Tell you whether this fits your investment thesis
  • Sense market timing
  • Suggest allocation calls

That’s still you.
This stack just clears the noise so your brain can focus.

Want the system as an AI Agent?

This prompt shows what AI is capable of.

But we didn’t stop there.

We built a full agent that takes this workflow to the next level, with deeper data, sharper outputs, and cleaner summaries. It runs the same process with more precision, adds custom logic for different deal types, and connects directly to your deal tracker.

It’s one of many tools we’ve built for our Pro Community.

💡
Access the Private Market DD Pro.