image-analysis

# Choom

A self-hosted AI companion framework with persistent memory, image generation, text-to-speech, and Signal messaging integration. Built with Next.js, Prisma, and a Python Signal bridge.

Each AI persona ("Choom") can have its own LLM model, voice, image style, and memory space. Chooms are agentic — they autonomously chain tools across multi-step tasks, searching the web, writing files, generating images, running code, and managing your calendar without manual prompting between steps. **Chooms can also delegate tasks to each other** — an orchestrator Choom can assign research to one agent, coding to another, and image analysis to a third, then synthesize the results. Talk to them through the web UI or via Signal messages.