What evidence shows an AI company has durable economic potential?

KR

Keith Rabois

Managing Director · Khosla Ventures

Compounding Data Loop

Keith Rabois says one of the survivors is a compounding data loop, along with data sliced by sub-vertical and corner-case depth in fields like law.

Itamar Novick

Founder & General Partner · Recursive Ventures

High Accuracy Metrics

Itamar Novick says AI buyers will soon compare systems by accuracy figures such as 99% versus 98%, which he argues will make it more obvious who the potential winner in a category is.

The answer starts at 19:07.

MF

Morgan Flager

Managing Partner · Silverton Partners

Proprietary Data

Morgan Flager says the biggest advantage he looks for is a proprietary data and learning advantage, since data others lack lets a company keep learning and sustain an edge competitors can't copy.

Nic Poulos

Founder & General Partner · Euclid

Proprietary Data

Nic Poulos says the first primitive of defensibility he looks for is data, including proprietary first-order data like Thomson Reuters' 40 years of annotated case law, which he says no one else likely has.

The answer starts at 12:18.

Garry Tan

Founder · Garry'S List

Founder’s AI Discussion

Garry Tan says investors get more signal from how a founder talks about AI in due diligence than from asking about model access, because the 2x-to-100x gap comes from context and workflow discipline and shows up in burn multiples before a pitch.