AI × Crypto Due Diligence Checklist for VCs

Lookout's DD framework for AI × Crypto, organized by stage.

Stage: first meeting (30 min)

Qualify before going deep. At the first meeting, only answer:

• Does the team have founder-market fit?
• Is the problem real and large enough?
• Is there any traction or early signal?
• Do I want to learn more?

If yes to all four → schedule a follow-up. Don't do deep DD at the first meeting.

Stage: second meeting — business DD

Team

• LinkedIn / on-chain history of all founders verified
• Previous company outcomes confirmed (not self-reported only)
• Reference calls with 2+ people who've worked with the founding team
• Cap table reviewed — founder equity, vesting, cliff issues
• Any co-founder conflict history? How was it resolved?

Product

• Live product demo (not slides) — request before the second meeting
• Smart contract audit status — auditor, scope, findings
• GitHub activity — commit frequency, contributor count, code quality
• On-chain activity verified via block explorer
• Any security incidents, exploits, or near-misses?

Market

• Customer interviews (minimum 3) — not founder-referred if possible
• Competitive landscape: who else raised for this, at what valuations?
• Regulatory exposure: jurisdiction, token classification, securities risk

Traction

• Raw data access (not dashboard screenshots) — Dune, SQL export, or API
• MoM cohort retention — not cumulative charts
• Revenue breakdown: protocol fees vs. token incentives vs. grants
• Churn rate and reason for churn

Stage: third meeting — technical DD

Bring a technical advisor or your engineering team.

For AI infrastructure

• Model architecture: proprietary vs. fine-tuned vs. API wrapper
• Data moat: what training data, how defensible, licensed?
• Inference costs: per transaction, margin at current and 10x scale
• Latency: acceptable for the use case vs. centralized alternatives?
• Fallback: what happens when the AI component fails or returns wrong output?

For AI agents specifically

• Agent autonomy level: what can it do without human approval?
• Error handling: how does it recover from failures? Who is liable?
• Identity layer: how is the agent identified on-chain? KYA compliance?
• Payment flow: how does the agent pay for services? Wallet management?

For DePIN

• Hardware provider incentive model: when/why do providers churn?
• Network utilization: % of capacity used vs. idle
• Geographic distribution: concentration risk?
• Physical infrastructure dependency: what if a key hardware supply disrupts?

Stage: term sheet — legal DD

• IP ownership: all IP assigned to the company? University / prior-employer claims?
• Token legal opinion: classification memo in relevant jurisdictions
• Employee / contractor agreements: IP assignment, non-compete, key-man
• Prior funding: SAFEs, convertibles, previous-round terms reviewed
• Any pending litigation or regulatory investigations?
• KYC / AML procedures for the protocol (if applicable)

Lookout's scoring framework

Maps to the 7-axis evaluation. Score each axis 1-5; weighted total out of 100 = the Lookout view.

| Axis | Weight | Key question | |---|---|---| | Team & Execution | 25% | Can this team ship and navigate adversity? | | Tech & Differentiation | 20% | Is the moat technical or just temporary? | | Tokenomics & Economics | 15% | Does value accrue to holders sustainably? | | Traction & Adoption | 20% | Real usage or incentivized metrics? | | Funding & Backers | 10% | Who's in the cap table? What signal? | | Narrative & Market Fit | 5% | Believable and timely? | | Risk Vectors | 5% | Are the main risks known and manageable? |

Lookout Library · updated 2026-05-28 · not financial advice