Lyle Jensen

@lylecodes

Seattle, WA

Joined September 2025

278

Following

Followers

Replies Media

Lyle Jensen
@lylecodes
Apr 5
OpenClaw vs Hermes Agent — quick comparison OpenClaw: Works across 50+ messaging apps, truly local tools by default, privacy-first assistant in your chats. Can involve more setup friction. Hermes (Nous): Strong self-improving loop, excellent persistent memory & auto-skills —
Lyle Jensen
@lylecodes
Apr 4
Anthropic’s 3rd party harness ban will definitely cost them subscriptions short-term. Heavy users who were squeezing max value out of Pro/Max are moving on. But this is probably a smart long-term move. They’re protecting their subsidized consumer tier from becoming a cheap
Lyle Jensen
@lylecodes
Apr 3
Your AI agents are blind without purpose-specific monitoring tools. We kept doing the same diagnostics manually — checking logs, querying the DB, tracing configs. So we turned each one into a CLI command. Debugging went from 20 minutes of grep to 5 seconds. If you're running
Lyle Jensen
@lylecodes
Mar 31
After running autonomous AI agent teams for days, here's the biggest lesson: AI is excellent at executing work. It's terrible at choosing work. Left alone, AI leads produce busywork — resurfacing empty tables, re-screenshotting unchanged dashboards, reassigning the same tasks.
Lyle Jensen
@lylecodes
Mar 30
Every AI agent framework lets you spin up agents. None of them answer the hard question: How do you actually manage a team of them? Who decides what they work on? How do you stop them from burning money on busywork? What happens when one goes off the rails? I've been running