I built a 17-agent AI swarm on my phone — here's how

Description and use

A 17‑agent AI swarm running locally on a Pixel 7 for workshop monitoring, built with TensorFlow Lite, Protobuf, and SQLite.

Problem

Run a meaningful AI swarm locally on a phone to process information and react intelligently, avoiding reliance on large cloud models.

Outcome

A functional 17‑agent AI swarm on a Pixel 7 that performs visual scene understanding and anomaly detection for workshop monitoring, demonstrating edge AI feasibility.

Workflow

• Collect hundreds of labeled workshop images
• Train object detection models with TensorFlow on desktop
• Convert models to TensorFlow Lite with post‑training quantization
• Implement 17 agents in Python, each with a dedicated TFLite model
• Use Protobuf for inter‑agent messaging
• Store shared knowledge in SQLite
• Activate agents selectively based on task needs
• Profile and optimize code with cProfile
• Deploy on Pixel 7 and run concurrently

Evidence

I built a system running 17 independent AI agents, all orchestrated on my Pixel 7.

Source

The agents aren’t all LLMs. That's key. I focused on targeted, specialized tasks.

Source

Running 17 agents concurrently on a phone requires serious optimization.

Source