Output surfaces

Docs are memory. Pages are how finished work ships.

The editor is still powerful, but it is agent-native: documents hold context the agent can read and update, while visual pages turn generated work into polished deliverables.

Markdown docs act as editable workspace memory

Inline AI edits can rewrite, expand, summarize, or restructure

Visual pages polish sections, layouts, embeds, and richer output

Same agent, files, tasks, and models across both surfaces

Agent memory surface

Document editor

For source notes, briefs, specs, research, and long-form outputs. The document is both editable work and memory the agent can reuse.

Highlight a section and ask the agent to improve it in context

Reference uploaded files and linked docs while editing

Keep the deliverable and agent memory in the same place

Finished output surface

Visual page editor

For emails, landing sections, reports, and pages where spacing, columns, embeds, and presentation matter after the agent creates the structure.

Turn generated drafts into shareable visual assets

Use when design hierarchy matters as much as the words

Keep the same thread, files, and model choices while changing canvas

Stay in a real document

Headings, lists, and blocks become durable memory the agent can use in future tasks.

Ask the agent on any section

Highlight content and ask for analysis, rewriting, expansion, summaries, or structured transformations.

Layout when the output matters

Switch to the visual editor when spacing, columns, embeds, or presentation quality affect the result.

From agent memory to share-ready output

Two surfaces, one thread, so the agent can keep context while you move from structured work to finished presentation.

Open workspace