Gobii Platform

Gobii Platform is an always-on autonomous AI Agent platform designed for teams and enterprises. Its core concept treats AI Agents as "Always-On AI employees" — virtual colleagues with independent communication identities, persistent memory, and real browser operation capabilities.

At the runtime level, each Agent has an independent scheduling state and a persistent event queue, triggerable via Celery Beat scheduling, Webhooks, or email/SMS events. SQLite operational memory provides a structured state substrate for long-running workflows, and native Agent-to-Agent communication enables multi-Agent coordination.

On the communication channel side, each Agent can be bound to a dedicated email address and SMS number. Team members interact with Agents through conventional communication methods without switching tools. Both inbound and outbound Webhooks are first-class operations.

On the execution side, the platform implements real browser automation based on browser-use (Playwright-driven), supporting headed browsers, persistent profiles, proxy-aware routing, and distributed Workers. Structured output is supported via the output_schema parameter with JSON Schema constraints, and MCP Server integration provides advanced extension capabilities.

For security, the platform offers encrypted keys at rest, proxy-controlled outbound traffic, sandboxed computation, and Kubernetes + gVisor backend isolation. Multiple LLM providers (OpenAI, Anthropic, OpenRouter, Fireworks, and custom endpoints) are unified through LiteLLM.

Deployment is Docker-based, supporting both self-hosted and Gobii Cloud managed modes. Functionality is extended on demand through multiple profiles (beat / email / obs) for scheduling, email listening, and observability. The open-source core uses the MIT license, while proprietary modes and non-MIT components require a commercial agreement.

Typical Scenarios:

• Revenue Operations Agent: Continuously monitors inboxes and web systems, auto-updates records and sends structured summaries
• Recruiting Operations Agent: Candidate sourcing, resume enrichment, persistent outbound message coordination
• Customer Success Agent: Inbound channel routing, browser operations, escalation with full state context
• Back-office Automation: Long-running, event-driven workflows requiring persistent memory and secure credential handling

Quick Start:

# Option 1: Managed installer
curl -fsSL https://gobii.ai/install.sh | bash

# Option 2: Manual clone
git clone https://github.com/gobii-ai/gobii-platform.git
cd gobii-platform
docker compose up --build

After launch, open http://localhost:8000 to complete initialization, create an admin account, select a model provider, add an API key, and create your first always-on Agent.

Architecture Overview: Monorepo containing agents/, api/, frontend/, billing/, sandbox_server/, console/, config/, middleware/, tasks/ modules. Web framework is Django (ASGI mode), task queue uses Celery + Redis, primary database is PostgreSQL, Agent state stored in SQLite, frontend is primarily TypeScript with Preline UI, package management uses pyproject.toml + uv.lock.