Venture Capital · Diligence & CTO-in-Residence · $6,000–$25,000
What a venture capital CTO does, and when a fund actually needs one
Three different jobs share this title, and funds routinely buy the wrong one. Written from a fund that deploys its own capital in crypto and AI infrastructure, so the diligence questions below are the ones we actually ask before wiring.
The fund's own internal CTO
Rare below roughly $250M AUM
An operating hire who runs the fund's own systems: deal flow, CRM, LP reporting, data pipelines, and increasingly the fund's own AI tooling. This is an internal operations job. It has almost nothing to do with judging whether a portfolio company's architecture will survive its next order of magnitude, and funds that hire one and expect the other are disappointed.
CTO-in-residence or technical venture partner
Usually fractional, often equity-linked
A senior engineer attached to the fund rather than to one company. Sits in on diligence, pressure-tests technical claims before the wire goes out, and is available to portfolio founders afterwards. Compensation is usually a retainer plus carry or a small allocation, because the value is judgement across the whole book rather than hours on one asset.
The portfolio-facing fractional CTO
Engaged per company, paid by the fund or the company
Placed into a specific portfolio company that has product-market fit and a non-technical founding team. This is a real operating role with a real backlog. It is the shape most likely to be confused with the other two, and the only one where the deliverable is shipped software rather than a written opinion.
Private equity
Audit an asset that exists
Venture
Underwrite a team's ability to build something that does not
PE diligence examines a running system with paying customers, a maintenance history, and a cost base you can measure. Venture diligence usually has a prototype, a roadmap, and three engineers. You are not scoring the artifact. You are scoring whether this specific team can produce the artifact on the capital being offered.
Private equity
Technical debt is a valuation input
Venture
Technical debt is often the correct choice
In a buyout, debt found in diligence becomes a remediation line in the model. At seed, a startup that has not accumulated debt has usually been building the wrong thing carefully. The question is not whether shortcuts were taken; it is whether the team knows which ones they took and what it costs to undo them.
Private equity
Key-person risk is a retention problem
Venture
Key-person risk is the entire thesis
A PE target can usually survive losing its lead engineer. A seed company frequently cannot, and pretending otherwise produces a diligence report that reads well and predicts nothing. Naming the two or three people the outcome actually depends on is more useful than any architecture diagram.
Private equity
Scale is measured against current load
Venture
Scale is measured against a load nobody has seen
The honest venture question is narrower and harder: at what specific point does this design stop working, roughly what does it cost to get past that point, and is that cost inside or outside the round being raised.
The buyout side of this is covered in more depth on the private equity CTO page, and the deliverable itself is specified on the technical due diligence service page.
Crypto and blockchain infrastructure
Fund I's mandate. The failure modes here are economic as often as they are technical, and a generalist diligence process misses them entirely.
- →Token and protocol design: what the token is actually for, whether the mechanism survives participants behaving adversarially rather than as modelled, and what happens to it in a market where nobody is speculating.
- →Validator and custody assumptions: who can halt the chain, who holds keys, what the recovery story is when a signer is unavailable, and whether the trust model in the deck matches the one in the code.
- →Upgrade and governance paths: whether a contract can be changed, by whom, and how quickly. An immutable contract with a bug and a mutable contract with an unaccountable admin key are different risks, not the same one.
- →Dependency surface: bridges, oracles, and sequencers a team does not control but whose failure is indistinguishable from their own.
AI and ML infrastructure
Fund II's mandate. Most AI diligence stops at the demo, which is precisely where the fantasy survives and the deployment does not.
- →Model and data dependency risk: what breaks if the underlying model provider changes pricing, deprecates a version, or ships the feature natively. A moat that assumes today's model is the last one is not a moat.
- →Inference cost curves: unit economics at ten times current volume, not current volume. Many AI products are gross-margin negative at scale and the deck shows the margin at demo scale.
- →Evaluation infrastructure: whether the team can actually tell when output quality regresses. A company that ships model changes without an eval set is flying without instruments and will not know it is off course until a customer says so.
- →Data rights and provenance: what they are permitted to train on, what they have actually trained on, and whether those two are the same sentence.
- →Moat durability against the next foundation-model release: the specific question of which parts of the product get absorbed if the frontier moves six months, and what is left that is genuinely theirs.
The Conflict Question
How does a fund manager run diligence for another fund?
Carefully, and in writing. Tykhe Ventures deploys $20M across two funds: Fund I from 2024 into blockchain and Web3 infrastructure globally, and Fund II from 2026 into AI-first companies in India as a SEBI AIF Category II.
If a diligence request sits inside either mandate, the answer is no, stated before any confidential material changes hands. That rule is simple enough to hold and specific enough to check.
Where the mandates do not overlap, the fund experience is the reason to hire rather than a problem to manage. Someone who has written cheques into infrastructure asks different questions than someone who has only read about it, and knows which claims in a deck tend not to survive contact with production.
Single-deal technical diligence
$8,000 – $25,000 per deal
Scoped to one company, delivered as a written report with a clear investment-relevant recommendation. Turnaround is typically 5 to 10 working days. Priced by deal complexity rather than cheque size.
CTO-in-residence, retained
$6,000 – $15,000 per month
Ongoing availability across the book: diligence on live deals, technical sanity checks for founders, and a second opinion on hiring senior engineers. Usually retainer plus a small carry or allocation, since the value compounds across the portfolio.
Portfolio company placement
$8,000 – $25,000 per month
A real fractional CTO seat inside a specific portfolio company, with a backlog and shipped software as the deliverable. Engaged when a company has traction and a non-technical founding team.
Got a deal you need a technical read on?
Thirty minutes, no pitch. If the deal sits inside a Tykhe mandate I will say so and decline before you share anything.