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

Description and use

A 17‑agent AI swarm running on a mid‑range Android phone autonomously gathers, analyzes, and synthesizes information on synthetic biology, producing a coherent report in about three hours.

Problem

Run a swarm of 17 independent AI agents on a mid‑range Android phone to autonomously gather, synthesize, and report on a complex topic without relying on a single monolithic model.

Outcome

The swarm produced a coherent summary report on synthetic biology research after ~3 hours, demonstrating that decentralized, specialized agents can outperform a single monolithic model on a phone.

Workflow

• Set up Termux and install Python
• Install LangChain, Ollama, SQLite libraries
• Configure Ollama with small models (mistral, llama2)
• Define 17 specialized agents (scrapers, summarizers, keyword extractors, contextualizers, researchers, fact checkers, synthesis agents, report generator) using LangChain tools
• Implement SQLite tables for agent memory and message passing
• Use LangChain Agent Executor to loop each agent’s decision process
• Run agents in controlled bursts to manage battery and RAM
• Monitor outputs and refine prompts to reduce hallucinations
• Compile final report via Report Generator

Evidence

I wanted to see if I could create a system that could autonomously gather information on a complex topic – let’s say, the current state of research on synthetic biology – and synthesize that information into a coherent summary.

Source

I chose smaller models like mistral and llama2 to ensure reasonable performance on my phone.

Source

After about 3 hours of running on my phone, the swarm produced a surprisingly coherent summary report on the state of synthetic biology research.

Source