Self-hosted AI chat for teams that remembers to charge your users

What it does

AChat is a self-hosted web platform that lets teams centralize AI conversations in one place. It wraps chat history, role-based access, content moderation, and usage billing into a single TypeScript codebase aimed at enterprises that want to keep data in-house. The project is currently iterating on V4; earlier versions were abandoned due to design flaws and scalability limits.

The interesting bit

Most open-source AI chat frontends stop at the conversation layer; this one includes commercial plumbing—payment integration and cost management—out of the box. It also claims to be “local-first” while offering remote synchronization, though the README never explains how that tension is resolved technically.

Key highlights

• Team-oriented RBAC and conversation management, not just a personal ChatGPT clone
• Built-in billing and payment tooling for commercial deployments
• Content moderation and data-compliance features for regulated environments
• Plugin system for workflow integration (details unspecified in the README)
• V4 is actively under development; V1–V3 are deprecated or frozen

Caveats

• V4 is still under development, and the project has a history of discarded versions (V2 was explicitly labeled “flawed design”)
• The README advertises “local-first” architecture and remote synchronization without clarifying the technical implementation
• Feature descriptions are high-level; specifics on plugin APIs, supported LLM providers, and deployment requirements live in external documentation

Verdict

Worth a look if you need to deploy a multi-user AI chat interface with usage tracking and basic governance. If you want a mature, stable foundation or a simple personal client, the version churn and enterprise complexity may not be worth it.