Tech
I think I have one of the best jobs in the world. I lead the technical side of the OpenAI startups team across EMEA & APAC, and that means that every week I get to see behind the scenes of the best AI startups globally.
Simultaneously, being at OpenAI, and embedded in the startups world, means I’m around incredibly smart people constantly thinking about the future.
I can only work with a limited number of companies directly, so this is a way to scale my learnings to more people. If I’m wrong on something, or you disagree on an opinion, please let me know. It will help me iterate to better beliefs.
-
Simple is powerful - but not everything is a coding problem
I've been thinking a lot about elegant ideas recently. Its a common finding in the sciences that elegant ideas tend to hold. Simultaneously, I've started hearing the phrase 'everything is a coding problem' come up in more and more settings.
-
Why even the best startups globally can end up with bad prompts, and how to fix this
1. Most prompts are bad because prompt evolution tends to be accretive: we only add, never remove, over time. This leads to spaghetti prompts, with contradictions and ambiguity. This has real business impact. 2. We need to treat prompt changes as product changes (because agent behaviour is product), and treat prompts as code (modularised, MECE, and all your other favourite acronyms; refactored if need be & actively maintained). 3. Well structured prompts enable teams to move faster, prevent regressions, and have better agents. I propose a very simple structure at the end.