If you are raising money for an AI-native company right now, investors are still using the same playbook — and most founders walk into that room with the wrong map. Brad Feld and Jason Mendelson wrote Venture Deals to fix that.
Feld and Mendelson spend the whole book insisting that founders overcomplicate term sheets because lawyers bill by the clause. Strip everything back and you are left with economics — liquidation preferences, participation rights, anti-dilution — and control — board seats, protective provisions, information rights. Every other term either flows from those two categories or exists to make someone feel important. Once you see the structure that clearly, you stop fearing the document and start negotiating the handful of things that genuinely affect your outcome.
An AI-native company has a wrinkle Feld and Mendelson could not fully anticipate: your leverage and your risk profile both look strange to a 2010-era term sheet. Agents compressing a twelve-person workflow into two humans and a stack means your burn is low, your headcount story is weird, and investors will reach for control provisions to compensate for their uncertainty about the model. Understanding which control terms actually bite — board composition above all else — lets you give ground on the theater and hold the line on what governs you for the next decade.
- Economics and control are the only terms worth a real fight — everything else is negotiating furniture
- low-burn AI-native teams unsettle investors, so expect compensatory grabs for board seats and protective provisions
- read the liquidation preference stack before you celebrate the headline valuation.
