Blog
Engineering, security, and product updates from the Archon team.
Introducing Archon v1.0
The production harness for AI agents is here. Cost control, security, observability, memory, eval, and governance — shipped as the framework itself.
How We Route 140+ Models for 65% Cost Savings
Our pattern-based router classifies complexity using 27 lexical signals in sub-millisecond time. No LLM call needed. Here's how it works.
Why API Keys Should Never Be Stored in Plain Text
We use the Stripe/GitHub pattern: SHA-256 hash on write, prefix+suffix for display, full key shown once. Here's why every SaaS should do this.
Four-Tier Memory: How Agents Remember
Working, episodic, semantic, procedural — inspired by cognitive science. With temporal decay and auto-consolidation to prevent context poisoning.
Building a Real-Time Dashboard with Convex
How we built a dashboard that updates in under 10ms using Convex reactive subscriptions, Clerk auth, and Recharts.