AI product, 2025

HomeAI

AI agent workspace | Full build | Next.js, Python, Postgres, Chakra UI

Internal product, architecture available for licensed builds

Multi-agent AI workspace with configurable models, tool integrations, workspaces, and persistent conversations. Local-first, self-hosted.

The problem

Running AI agents meant bouncing between ChatGPT, API playgrounds, and scripts. No way to create purpose-built agents with specific tools, save conversations, or share workspaces across devices.

What I built

Full workspace application: create agents with custom system prompts and model settings, attach tools and MCP adapters, organize conversations by workspace, and run everything on your own hardware. Python backend handles model routing and tool execution.

Results

Multi-model support: swap between GPT, Claude, and local models per agent
Tool and MCP adapter system for extending agent capabilities
Workspace isolation: separate contexts for personal, work, and client projects