one thing i have found remarkable in both scientific discovery and quant trading is simply knowing a competitor is close to a solution is enough to find an answer yourself. i saw it first-hand many times in CRISPR and there are myriad examples throughout history (4 minute mile, soviet bomb after trinity, explorer after sputnik, deepseek r1 after o1)

Turns out, switching to grok bot was a huge mistake (sorry @jessethanley) Experience-wise, nothing comes close. I HATED configuring my openclaw/hermes agents and fixing what was broken. In contrast, Grok Bot literally JUST WORKS™ It's a joy to use, spinning up new agents is a joy, and it feels infinitely powerful. But right now, the worst thing about grok bot is grok 😅 Grok 4.6 just isn't that good of a model for tasks outside coding. Its communication style + intuition are _terrible_ I'm using it for triaging support tickets, drafting replies, investigating bugs, managing my outbound etc. But in @bot, it was making mistakes that even open models like DS V4 Flash would never make... - malformed tool calls - failing to sum two numbers together (!) - and just a broad misunderstanding of my intent In practice, Grok was far worse than just using codex or claude. I tried all I could to steer it/get it to work better, but the model just isn't there yet I think. So canceled my grok sub and moved back to my hermes + codex + telegram setup. After tasting the grok bot onboarding, doing this was excruciatingly painful. Hoping desperately that the @ChatGPT team do something directionally similar very soon, as with Astra, I'm this 🤏 close to canceling Claude