Tetriz Engineering Intelligence

Tetriz Engineering Intelligence is an open-source Go library from Workloom Solutions, published as tetriz-ai, for analysing engineering team data: AI coding session logs, PRs, issues, diffs and branches. The repository has 10 packages, and 9 of them run fixed-rule analysis offline on your machine. The tenth, prworktype, classifies PRs as bug fixes, features or other work, in layers: rules, then local MiniLM embeddings, then an optional large language model, and it abstains when it isn't confident. What sets it apart is the clear split between the AI part and the plain statistics, so a team can use only the offline packages. The contributors file says AI coding agents including Claude Code, Codex, Cursor, Gemini and Kiro were used on the code.

Who it's for:Engineering leads and platform engineers who want to measure how AI-written code is working out for their team. If you don't write code, you won't need it.

Before you start: You need Go 1.25.2 or later, and you wire it into your own program as a library; there is no ready-made dashboard. prworktype also needs Python 3.10 or later, or Docker, and first-time setup pulls about 1 GB of dependencies, which the authors estimate takes 10 to 60 minutes. With LLM classification on, PR file lists and descriptions go to whatever model service you configure (a local Ollama, or cloud services such as OpenAI or Gemini), paid from your own account. The AIDLC classifier sends item names and descriptions to Gemini; the authors redact sensitive patterns first but say this cannot guarantee nothing leaks. Company code and employee session logs are sensitive, and the statistics and sample ROI figures should not be treated as a verdict on anyone's ability. The code is open source under the MIT licence.

Based on the project's public documentation. Not tested by AI Inspiration World.