P1 and agentic usage

P1 provides a governed way for people and AI agents to create and evolve web experiences.

An agent is software that can interpret a goal, use available tools and context, and complete a sequence of steps. In P1, agents can participate in web work alongside people rather than operating in isolation. They can use the same business context, approved components, permissions, and workflows that guide other contributors.

What P1 provides

P1 is the layer around agent-assisted web work. It helps connect an agent’s work to:

• The site’s business and content context
• Approved components and design-system rules
• User and project permissions
• Collaboration, review, and versioning
• A controlled path toward testing and production

This makes agentic usage more than generating code or content. The result should be visible, reviewable, attributable, and reversible before it affects a production experience.

People remain in control

P1 does not assume that agents replace web professionals or make every decision autonomously. Agents can increase the number of contributors and the amount of work that happens in parallel, while people remain responsible for consequential changes and approvals.

A useful mental model is a multiplayer web team: people and agents work from shared context, and P1 supplies the workflow needed to inspect, discuss, test, approve, release, or roll back changes.

Agentic Interaction with P1

In addition to using one or more agents inside the P1 editor, P1 can be connected to your preferred tools via MCP. Learn more about agentic control of P1 here.