Introduction

Padiso is an agent orchestration and unified integrations platform. Instead of running AI agents by hand, you describe an outcome, hire a team of always-on agents, connect the tools they need through the Model Context Protocol (MCP), and let them run your operations, with budgets, human approvals, and a full audit trail.

What Padiso does

• Orchestrates a team of agents. Agents run as background workers on a heartbeat, advancing their tasks, coordinating with teammates, and reporting outcomes, without you babysitting a terminal.
• Connects everything via MCP. Connect a tool, API, database, or model once and every agent in the workspace can use it, scoped and governed.
• Runs anywhere. Use the desktop app on your own machine, or deploy to your cloud (AWS, Azure, GCP).
• Keeps you in control. Per-agent budgets, human-in-the-loop approvals, and an audit trail on every action.

Core concepts

• Workspace, your isolated environment: agents, projects, tasks, runs, connections, and spend.
• Agent, a worker with a role, a runtime (model/engine), goals, and a budget.
• Goal & tasks, what an agent is working toward, broken into tasks it advances each cycle.
• Run, a single execution of an agent on a heartbeat. Runs are persisted, deduped, and retried.
• Integration (MCP), a tool, API, or model exposed to agents through the Model Context Protocol.
• Approval, a gate a human clears before a sensitive or irreversible action proceeds.

Where to go next

• Quickstart, set up your workspace and hire your first agent.
• Concepts, agents, goals, tasks, runs, and the heartbeat.
• Integrations & MCP, connect tools, data, and models.
• Runtimes & models, choose an engine per agent.
• Governance, budgets, approvals, and audit.
• Deployment, run locally or in your cloud.