Prediction: OpenAI chases personal assistants while Anthropic defends coding
The author bets OpenAI will shift to the application layer and dominate Muse/Grokbot/Instinct-style personal assistants, while Anthropic keeps making best-in-class models but may lose out by staying mostly a model company.
My bet is that OpenAI will shift towards the application layer and focus on dominating the Muse/Grokbot/Instinct personal assistant, and everyone will have one (eventually hardware as well) While Anthropic will stay the king of coding, world's best-in-class models And then you will have your bots coding with claude as well (they have their own VM so you can just pass whatever harness and model you want) I think this can actually be worse for Anthropic because at this point they are mostly a model company. It's way better to own the distribution layer like Meta or OpenAI. If a new open source model get's better or very close to Fable, or if Astra also get's closer, it's way less friction to change a model, then to change a whole personalized AI assistant setup custom to your tools (akin to setting up a new openclaw all over again)