Portia AI

Portia AI is an open-source framework for building AI agents that require human oversight at critical decision points. Unlike fully autonomous agent frameworks, Portia takes a human-centric approach: agents execute automatically where confidence is high but pause and request human input when they encounter ambiguity, insufficient permissions, or high-stakes decisions — creating a collaborative human-AI workflow.

The framework introduces "clarifications" as a first-class concept: structured requests from the agent to the human for additional information, approval, or guidance. This makes agents significantly more reliable in production because they don't silently make wrong assumptions or take irreversible actions without appropriate review. The result is AI that is both more capable (able to handle complex tasks) and more trustworthy.

Portia integrates with all major LLM providers and the growing ecosystem of MCP (Model Context Protocol) servers for tool access. Its clear separation of agent planning, tool execution, and human interaction layers makes it easy to build, test, and debug complex agent workflows.