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

Description and use

A 17‑agent swarm runs locally on an Android phone, with sensors, processors, and actuators communicating via ZeroMQ, built using Python and Termux.

Problem

Running complex AI tasks on a phone is hard due to limited memory, constrained processing power, and unreliable background execution.

Outcome

The phone successfully runs a 17‑agent swarm that simulates sensor data, processes it, and triggers actuator actions, demonstrating that mobile devices can host decentralized AI systems.

Workflow

• Install Termux and Python on the phone
• Set up ZeroMQ for inter‑process messaging
• Write sensor agents that publish simulated data
• Write processor agents that subscribe, process, and decide actions
• Write actuator agents that subscribe to processor outputs
• Run each agent as a separate Python process in Termux

Evidence

Running complex AI tasks on a phone presents some hurdles. Memory is limited, processing power is constrained (compared to a server), and background execution can be… unreliable.

Source

My tech stack ended up being surprisingly straightforward: Python, Termux, ZeroMQ, Simple AI Algorithms

Source

The swarm consists of 17 agents. Each agent has a specific 'role' and communicates with the others via ZeroMQ.

Source