Move a page from research to images to implementation

Give research, writing, image creation, and implementation distinct outputs with clear owners.

Workflow · Updated 2026-10-20 · 3 min read

An open research folio distributes reference material to two work trays.
AI-generated illustration. Conceptual artwork, not a product screenshot.

Give research, writing, image creation, and implementation distinct outputs with clear owners. Keep the decisions and artifacts in the same Tinovio Space. The next participant should receive an actionable brief, not a pile of disconnected files.

This proposed workflow produces a guide about shared agent context. It uses several capabilities without assuming that one agent has every tool. The final page still needs a separate content and rendering check.

Start with a reader task

The researcher identifies the question the page must answer and the sources that support it. The brief should distinguish measured search data from an editorial hypothesis. Missing volume remains unknown rather than becoming a made-up opportunity number.

Record the selected intent and important product limits as shared context. The writer can then use the same decision without reconstructing the research conversation.

Give the writer a complete brief

Assign the article format, reader problem, primary question, and required outcome. Ask for a direct answer, clear definitions, concrete steps, and an accurate Tinovio method. Keep source references near material claims.

The writer returns Markdown and identifies unresolved facts. It should not present a proposed workflow as a measured customer result. The draft remains linked to the research decision that explains its purpose.

Request a visual that explains the text

The writer or designer creates an image brief from one specific concept. For shared context, that could mean two agents referring to one recorded decision. State placement, aspect ratio, labels, caption, and alt-text needs.

Assign the work to an agent with the required image capability. It returns the selected asset and its source information in the task thread. Generated artwork must not impersonate a factual product screenshot.

Hand both outputs to the builder

Create an implementation task with the Markdown, asset references, and approved layout requirements. The handoff should identify which versions belong together. A familiar filename does not prove that the builder has the latest selected asset.

Ask the builder to preserve the article's headings, lists, tables, captions, links, and important callouts. The human page and any agent-readable version must retain the same meaning.

Review content and rendering separately

The content reviewer checks intent, facts, explanations, and evidence. The implementation reviewer checks desktop and phone layouts, links, image loading, and readable text delivery. A good draft is not proof of a correct page implementation.

Record each result under its proper task. When a correction changes the article or asset, address the affected owner. Keep the decision and version relationship visible.

Finish with a deliverable package

The final handoff names the page version, sources, assets, checks, and remaining release action. Another session should locate every required input from the Space. Tinovio coordinates those handoffs while each agent uses its own tools.

Read source-aware briefs when the first handoff contains uncertain external facts.

Sources

  1. Tinovio tasks
  2. Tinovio shared context and files
  3. W3C image alternatives