How Dollie Uses Editor

Dollie is an Agents-as-a-Service platform. Operators use it to create and run agent-powered products for their customers.

Those products need application surfaces around the agents: landing pages, service pages, course pages, onboarding experiences, and other customer-owned content. The pages must be flexible enough for each customer while remaining consistent with the product that hosts them.

The original requirement

Dollie needed a way to let customers customize pages without giving them an unrestricted website builder.

The system had to support:

• application-owned React components;
• multiple component libraries without id collisions;
• tenant-owned Page definitions;
• live backend data;
• localized page variants;
• human editing and agent-assisted drafting;
• server-side validation before persistence;
• published rendering from the same components used in the editor.

Why a catalog

A catalog lets the host application define its page vocabulary.

Instead of telling an agent to “write a landing page,” Dollie can tell it which heroes, feature sections, pricing layouts, illustrations, and calls to action exist. The agent fetches the fields for the components it needs and returns a structured proposal.

The operator or customer can review that proposal in the visual editor. The application validates it before saving.

This keeps agent output inside the product's visual and operational boundaries.

Why the SDK is open source

The same requirement appears in many SaaS products.

A course platform wants instructors to customize sales pages. A vertical SaaS wants each tenant to maintain a branded landing page. A consultancy wants clients to edit approved sections without breaking the design system.

These teams should not need to build the same catalog, editor, validation, persistence, and Copilot infrastructure from scratch.

Dollie Editor is open source because the underlying machinery is useful independently of the Dollie platform.

Dollie is proof, not a dependency

Dollie remains the flagship production user, but the SDK does not require:

• a Dollie deployment;
• Laravel;
• the Dollie agent runtime;
• a particular AI provider;
• Dollie's storage or tenant model.

Those integrations demonstrate one way to use the contracts. Your application keeps its own architecture.

The rule that carries across both products

The Copilot proposes. The user decides.

Dollie Editor gives AI enough structure to be useful without giving it a second, ungoverned path around the application.