← Back to Home

J&M Labs Current Status

Last updated July 7, 2026

Scope: This is the public-site orientation page. It summarizes what the published pages currently point readers toward; it is not a private fleet health dashboard and does not expose live internal endpoints.

Current architecture

Start with the fleet architecture snapshot and current benchmark page before reading older experiments.

Local LLM Stack

Milo / Hermes

Hermes migration is treated as profile-separated shadow/cutover work, not identity blending.

Milo Migrating to Hermes?

Memory

Honcho is the public current memory direction; OB1 is described as archive/provenance unless a page says otherwise.

OB1 to Honcho checklist

Benchmarks

Benchmark numbers are dated operational measurements. Keep harness, engine, timeout, host, and context settings separate.

Benchmark topic page

Current public entry points

QuestionStart hereWhy
What is true about the local AI stack?Local LLM StackCurrent architecture and benchmark orientation.
What changed in memory?OB1 to HonchoMigration state, continuity repair, and recall framing.
What should agents read first?llms.txtMachine-readable site purpose, reading rules, and high-value links.
What is newest?RSS feedRecent dated posts in a simple machine-readable feed.

Recent public updates