SuperAPI AI Directory

gobii-platform

Gobii Platform is an open-source AI employee platform for running persistent, autonomous agents in production environments. Each agent can run continuously, wake up on a schedule or event, use a real browser, call external systems, and collaborate with other agents. Agents also have their own identity (such as email or SMS) and can be contacted like employees, enabling 24/7 work. This repository is intended for self-hosting and development, and the hosted product is Gobii Cloud. The platform emphasizes reliable, secure, always-on team-level agent operations, supports encrypted keys, proxy egress control, sandboxed compute, and provides integration methods such as API, Webhook, and remote MCP.

Open website

About this tool

Gobii Platform is Gobii's open-source platform, intended for self-hosting and development, and the hosted product is Gobii Cloud (gobii.ai). It is positioned as an "AI employee" platform that can run persistent, autonomous agents in production environments.

Purpose: For teams with real work to do, letting agents work continuously like employees. Each agent can run continuously, wake up on a schedule or event, use a real browser, call external systems, and collaborate with other agents. Agents can be assigned identities (email or SMS), be contacted directly, and keep working 24/7.

Main capabilities:

  • Always on by default: each agent has scheduling state and supports persistent event handling.
  • Agents with identity: each agent can have its own email address and SMS phone number, and teams can contact it directly.
  • Native inter-agent messaging: linked agents can coordinate directly.
  • Webhook-native integration: inbound webhooks wake agents, and outbound webhooks are a first-class agent action.
  • Based on browser-use: maintains /api/v1/tasks/browser-use/ compatibility and adds platform-level runtime controls.
  • SQLite-native operational memory: provides a structured state foundation for long-term tool workflows.
  • Real browser operation: headed execution, persistent profile handling, proxy-aware routing.
  • Security-first controls: encrypted secrets at rest, proxy-managed egress, support for Kubernetes sandboxed compute.
  • Remote MCP: Gobii Cloud is a hosted MCP server, and clients can create, manage, message, coordinate, and inspect persistent agents.

How to use: Self-hosted installation requirements: Docker Desktop (or a compatible engine), with at least 12 GB RAM allocated. Run the hosted installer, supporting GOBII_INSTALL_DIR (default ~/gobii-platform) and GOBII_REF (default latest tagged release). Stable installation endpoint:https://gobii.ai/install.sh。安装后打开 http://localhost:8000 Complete setup: create an administrator account, choose a model provider (OpenAI, OpenRouter, Anthropic, Fireworks, or a custom endpoint), add API keys and model configuration, and create your first always-on agent.

API and integrations: provides an API quick start. Remote MCP endpoint:https://gobii.ai/api/v1/mcp/,认证使用 Authorization: Bearer <key> or X-Api-Key: <key>. Currently provides 16 tools and authentication file resources. Self-hosted deployments can replace the endpoint with http://localhost:8000/api/v1/mcp/。

Applicable scenarios: revenue operations agents, recruiting operations agents, support and success agents, back-office automation, and other long-running trigger-driven workflows that require persistent memory and secure credential handling.

Limitations:

  • Docker must have at least 12 GB RAM.
  • Self-hosting requires completing configuration and management yourself.
  • Hosted Gobii Cloud is available, but open-source self-hosting is intended for self-hosting and development.
  • Remote MCP requires an API key, and keys must not be committed or pasted into chat.
  • The source code is MIT licensed, and the Gobii name and logo are trademarks; proprietary modes and non-MIT components require a commercial agreement with Gobii, Inc.
  • Specific information such as fees and pricing is not provided in the documentation.