Pinned
Agents can fetch data and nail the facts, but combining it all into real research is where they start to break.
Sentient researchers Darshan Tank (@TankDarshan7) and Sidhant Rahi tackle this in "CryptoAnalystBench: Failures in Multi-Tool Long-Form LLM Analysis", accepted into
