Korva is a scroll of the internet’s best ideas.
Labs build small products to win teams
Startups that bring AI to a market by doing the right integrations and customizing the system prompt are fine as long as the model labs don't care about that market, and plenty of markets are like that. Labs have several incentives to go after a company eventually.
One incentive is getting multiple teams inside the companies they care about to depend on them. That is one theory for why Claude Design was strategic: it probably isn't a significant amount of revenue for Anthropic, yet it gets the design team to really care about Anthropic models. The company then has another team that wants to stick with Anthropic.
A startup that builds a product for a team that has become strategic to the labs sits closest to the threat.
Why does one team's adoption matter so much?
Software vendors often start with a single team and spread across a company, a strategy known as land and expand. Once a team's daily work runs through a vendor's tools, moving to another vendor means retraining people and rebuilding workflows. Each additional team that adopts the tool adds to the cost of leaving.
How have platforms absorbed startup features before?
The tech industry has a word for it: getting Sherlocked. It comes from Apple's Sherlock search tool, whose third version in 2002 added features that closely resembled a third-party app called Watson from Karelia Software. Since then the word has been used for any platform that folds a popular add-on into its own product.
More from 20VC with Harry Stebbings
