@jointhebridge We are building specialised teammates that will increase the value per dollar of your llm.
Co-operative agents that work autonomously with proper guarding, permissions... and many funnier things like AG-UI for your reports.
Ok ok, got an even more controversial one. In THEORY yes that's true, but that's not taking into account the "caching" component, as routing on a cold smaller model might cost more than a heavy hot model with cache.
So, benchmarks should take it into account, and benefits will be weak.
@sridharfyi Yes ! Coordination is key, we found our engine to give 5-7x better output with multiple specialists (benchmark on divergence of ideas, actionability, relevance) per dollar spent on tokens, compared to Claude cowork.
PRO TIP : ask @claudeai to estimate time not in hours (it's literally useless, it will be done in minutes anyway) but in tokens and value. Prompt in comment 👇
@jointhebridge Hey ! Would definitely love to join
Building https://t.co/TrJH0fyZQ5 the easiest way to build a team of AI specialists to accomplish your boring tasks. Happy to send a demo !
@dessaigne This is going to be crazy, hopefully YC spirit is fungible with Paris!
Still on the sideline, but definitely hoping for a mail. @dessaigne wish me luck 😸
Building unique AI Teammates on https://t.co/TrJH0fyZQ5 and the most accessible AI companion with https://t.co/VUBgyYuTfw