Jul 7, 2026
Your Agent Isn’t Dumb. It’s Just Lost.
A per-window Table of Contents in the MCP tool layer, plus an Auto Atlas for long-term context, so your agent stops wasting its context window hunting for relevant information.
Latest engineering posts
Your Agent Isn’t Dumb. It’s Just Lost.
Jul 7, 2026
A per-window Table of Contents in the MCP tool layer, plus an Auto Atlas for long-term context, so your agent stops wasting its context window hunting for relevant information.
sessiongrep: A Local-First Memory Layer for CLI Agents
May 27, 2026
SQLite + FTS5 turns raw Claude Code and Codex CLI transcripts into something humans and agents can actually search.
Deployed at Brain Co.
May 14, 2026
Hear from two deployed engineers about their day-to-day innovating AI products with Brain Co.
An AI Blitz: Highlights from Brain Co.'s Hackathon 2026
Mar 20, 2026
Featured projects from our company-wide AI hackathon.
All engineering posts
May 27, 2026
sessiongrep: A Local-First Memory Layer for CLI Agents
SQLite + FTS5 turns raw Claude Code and Codex CLI transcripts into something humans and agents can actually search.
May 14, 2026
Deployed at Brain Co.
Hear from two deployed engineers about their day-to-day innovating AI products with Brain Co.
Mar 20, 2026
An AI Blitz: Highlights from Brain Co.'s Hackathon 2026
Featured projects from our company-wide AI hackathon.
Feb 11, 2026
Confidence Scoring with Bayesian Networks
Brain Co. uses a Bayesian Network approach to confidence scoring that enables full automation of the building permit approval process (Figure 1).
Dec 19, 2025
LLM-Generated Rules Engines: Executable IF-THEN Logic for LLM Explainability in Regulated Industries
A technique to make LLM decision making auditable when processing unstructured documents.