Building without dependencies

Was this email forwarded to you? Sign up here

OpenAI has gone from the clear winner to a complete question mark. What the future looks like nobody knows. For all the companies building exclusively on OpenAI, this is the wake-up call.

Over the years I have seen countless companies building exclusively on someone else’s datasets. There was the ecosystem of companies built on Twitter dating back the the GNIP years. Then I experienced the dozens of companies that were building on top of Yodlee to provide consumer transaction data insights to hedge funds. In a similar, but different approach, Bloomberg for many years made their order management systems free for Bloomberg Terminal users. Then once all the companies using it depended on it, they flipped the switch and went from free to seven-figure deals.

Building on top of another company is dangerous. If we have learned anything over the years, it has been reinforced this weekend, diversify. Have redundancy. Build on multiple models. Build with multiple datasets. Having your entire business rely on one model provider or one core dataset, which you don’t ultimately own is dangerous.