Blog
Technical deep-dives, architecture decisions, and war stories from building an autonomous agent runtime.
Squad: Open-Source Multi-Agent Orchestration
Gateway-centric, connector-based, Docker-first. Squad is an open-source platform for running coordinated multi-agent workflows in production. Built to replace brittle script-based orchestration with a proper abstraction layer.
Read post →Private Notes, Local AI
Every note-taking tool I tried either stored my data on someone's cloud or charged per AI query. Jot captures notes to SQLite instantly, then enriches them in the background with a local model: tags, action items, linked notes. No API keys, no cloud, no per-query cost.
Read post →Building a Collaborator, Not a Chatbot
Lobs Companion is a macOS app that runs entirely locally. Three memory layers (session, project, permanent), a Unix socket bridge between Tauri and Node.js, and a glass morphism UI built to feel native. A collaborator that already knows your codebase.
Read post →Extracting a Toolkit from Production Code
I built lobs-core, then started a new project and found myself copying the same patterns from scratch. Agentic is five TypeScript packages extracted from that production code: LLM client, tool executor, memory, runner, and config. About 9,000 lines, all battle-tested before becoming a library.
Read post →One Agent Per Course
CourseClaw gives each university course its own OpenClaw agent in a dedicated Docker container. Before any architecture decisions came four hard rules that nothing can override — starting with: student messages never trigger writes to course material.
Read post →The Structured-First Executive Assistant
Most AI productivity tools are chatbots with a calendar wrapper. Cortex is different: tasks, events, and projects live in Postgres as first-class objects, and four separate AI roles reason over that structure — proactively, on a schedule, without being asked.
Read post →Preparation Beats Power
Most of the improvements I've made to Lobs in the last week aren't about making the agent smarter. They're about making it less wasteful.
Read post →Teaching an Agent to Listen
Lobs could read, write, and search — but only through text. Adding a Discord voice pipeline with local STT/TTS, a realtime mode for sub-second latency, and live meeting transcription with an AI activity feed that extracts action items while you're still talking.
Read post →Why I Built My Own Agent Runtime
Every version of Lobs was forced into existence by hitting a hard ceiling in the previous one. What started as a chat plugin became a standalone runtime with 6 specialized agents, a workflow engine, and 47K lines of TypeScript. Here's why, at each step, the only move was to go deeper.
Read post →The Restart Loop: When Your AI Agents Go Rogue
Workers edited the runtime source code, then called restart. The restart spawned fresh workers. Who picked up tasks. Who edited source and called restart. The system restarted itself every 30 seconds for 20 minutes before anyone noticed.
Read post →Five Tiers, One Rule: The Cheapest Model That Works
Running AI agents 24/7 on a grad student budget means every token matters. Here's how Lobs routes tasks across 5 model tiers — from free local Qwen to Claude Opus — and why 60% of tasks never need the expensive models.
Read post →