Introducing EM-Aide: an agentic AI copilot for Engineering Managers
Over the past few weeks, I’ve been building EM-Aide, an open-source, local-first agentic AI system designed to help Engineering Managers answer a very practical question:
“Based on how my team is actually working, what should I focus on this week?”
EM-Aide ingests real delivery signals from GitHub (and Jira where applicable), computes actionable metrics, and generates a weekly EM brief with:
- Prioritized actions
- Key risks to watch
- Clear evidence behind every recommendation
All without sending code, diffs, PR comments, or ticket text to an LLM.
🌐 Live demo (working)
👉 https://em-aide.annvay.com
What makes EM-Aide different
🧠 Agentic, not chat-based
This isn’t a chatbot. It’s a structured pipeline: ingest → metrics → context → plan. Inputs are bounded, outputs are repeatable, and every run is explainable.
🔒 Privacy & auditability by design
Only sanitized, aggregated metadata is sent to the model. You can even preview the exact data payload sent to the LLM (not the prompt), which makes the system far more trustworthy.
🖥️ Local LLM support (Ollama)
EM-Aide can run fully locally using Ollama, or switch to a hosted LLM via configuration. One-command Docker startup keeps setup friction low.
📦 Operationally simple
Backend, worker, UI, database, and optional local LLM all start together. Clean resets make experimentation easy.
🌐 Enterprise-aware
Supports GitHub Cloud and GitHub Enterprise, correct PR linking, and is architected for multiple repos per team.
Why I built it
Most EM tools either overwhelm you with dashboards or provide generic AI advice with no grounding in reality. EM-Aide aims to sit in the middle:
Opinionated recommendations, grounded in delivery data, with strong guardrails.
The project is open source and evolving:
👉 https://github.com/hemchand/em-aide
If you’re an Engineering Manager, Staff+ engineer, or tech leader curious about how agentic AI can support real engineering leadership, I’d love feedback after you try the demo.
#EngineeringManagement #AgenticAI #OpenSource #DeveloperProductivity #TechLeadership