Tool Use - AI Conversations Hermes Agent has won. Here's why
42 snips
Apr 14, 2026 
Robert Desmond, founder of Farm Friend who builds non-technical AI workflows; Wolfram Ravenwolf, AI evangelist and Wolfbench creator who benchmarks model/agent performance; Karan, Nous Research co-founder leading model behavior work. They discuss why Hermes Agent is winning, real-world automations like video dubbing and CAD sourcing, Hermes’ self-improving skills and three-layer memory, and the case for open, switchable AI infrastructure.
AI Snips
Chapters
Transcript
Episode notes
Ambient Agents Are An Intelligence Flywheel
- Ambient agents replace chatbots by proactively working in your context instead of waiting for prompts.
- Mike Bird contrasts chatbots as gateways and ambient agents as an “intelligence flywheel” that optimizes schedules, research, and recurring tasks.
Let Your Agent Autonomously Manage Memory
- Let the agent autonomously manage memory while you inspect or edit when needed.
- Hermes uses curated short markdown files, consolidation, session search, and autonomous pruning to keep memory useful and stable.
Auto-Created Skills Turn Repetition Into Reliable Playbooks
- Skills let agents follow repeatable playbooks and Hermes can auto-create and update skills in the background.
- Robert Desmond and others highlight a sub-agent that analyzes sessions and turns recurring actions into skills without interrupting you.