Vibesboard docs#

Vibesboard is a self-hosted, multi-tenant platform for building and operating customer-facing AI agents. It combines the agent runtime with the parts teams usually have to stitch together around it: knowledge and memory, messaging channels and inboxes, business tools, scheduling, model routing, access control, and usage management.

Quickstart

Run Vibesboard locally with Docker Compose in a few minutes.

Core concepts

Workspaces, agents, knowledge, and memory — how they fit together.

Build

Ground agents in your data and connect them to tools.

Deploy

Ship an agent to your website, WhatsApp, or Instagram.

Self-host

Run the full stack yourself, or deploy it to Cloud Run.

Platform

Multi-tenancy, bring-your-own-LLM, and the security model.

Why Vibesboard#

Many tools make it easy to demo a chatbot. Vibesboard is built for the operational work that starts after the demo: agents where customers already are, answers that lead to outcomes, one deployment for many teams or clients, and freedom from model lock-in.

Open core · public beta

Vibesboard's core is MIT licensed and self-hosted. The project is pre-1.0, so read the open-core boundary and run it on your own infrastructure before opening an issue if something's missing.