New day, new blog post on the nature of agents. blog.exe.dev/topology↗

Introducing AgentMail Evals. We tested how AI agents use tools: 25 model and harness combinations, seven email tasks, with AgentMail as the proxy.↗

Claude Opus 5.5 is the best LLM for Operations we've benchmarked. It scored 40% on @Zapier's AutomationBench: 657 of the hardest real business workflows we have, across 6 departments. It excelled at Ops work where the model has to go find the procedure and use several tools ac…↗

Really excited about this post. We wanted to understand an agent harness from first principles and share our findings publicly. No bullshit, no AI generated slop, just pure alpha. rubriclabs.com/blog/what-is-a…↗

By now this is clear to many of us, but it is worth writing down. The age of “no code” has passed. blog.exe.dev/the-end-of-no-…↗

Benchmarks are dead (for us). Our RSI system makes benchmarks too easy. Give ours a benchmark and it builds its own solution–then beats SOTA. We just did it on 6 at once: math, coding, planning, long-context, tool use, web apps. No human tuning.↗

Introducing Muse Spark, the first in the Muse family of models developed by Meta Superintelligence Labs. Muse Spark is a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration. Muse Spark is available today at …↗