When your company's core labor is done by agents, investors will price the cap table differently — and you need to know exactly which clauses let them reach in and adjust the controls.
Brad Feld and Jason Mendelson wrote Venture Deals as a map through a process that deliberately obscures itself. Their central argument is surgical: strip any term sheet down to its bones and you find exactly two things worth fighting over, economics and control. Liquidation preferences, participation rights, and anti-dilution provisions determine who gets paid and how much. Board seats, protective provisions, and drag-along rights determine who can say no. Every other clause is mostly noise that keeps lawyers billing.
An AI-native company changes the leverage calculus at the table in ways Feld and Mendelson could not have anticipated but their framework handles cleanly. If your agents are the primary value creators, a sophisticated investor will push harder on protective provisions — because the humans holding judgment are fewer and the system is harder to inspect. Control clauses that once felt boilerplate now govern whether you can retrain a model, swap an infrastructure provider, or pivot an agent's scope without triggering a board vote. Know which boxes on that term sheet are actually levers before anyone slides it across the table.
- Economics and control are the only two variables worth negotiating — locate them fast and ignore the theater around them
- When agents generate the output, investors will treat protective provisions as operational controls, not just governance formalities
- The fewer humans in your production layer, the more a counterparty will want contractual handholds — price that into your redlines early.
