Poppy field guide · 02
Research-to-content workflow
A repeatable route from collected sources to publishable material without flattening uncertainty or losing provenance.
Tutorials, reviews, briefs, newsletters, and scripts that must remain tied to a body of source material.
A smooth synthesis turns mixed evidence into one confident story and hides what was inferred.
Start condition
Define the editorial promise before collecting the universe.
Research expands to fill the available board. A clear promise limits that expansion: the finished piece will help a named reader make a decision, complete a workflow, or avoid a failure. The promise should also state what the piece will not claim. That prevents a practical tutorial from quietly becoming an exhaustive product review it cannot support.
For Top AI Tools For, the strongest promise is usually operational. Explain the job, show where the tool fits, identify what breaks, and leave the reader with a decision or procedure they can use. A long feature inventory is not evidence of depth when it never becomes a working recommendation.
- Write the reader, decision, evidence threshold, and excluded claims at the top of the board.
- Identify the one demonstration or artifact that would make the piece defensible.
- Stop collection when new sources repeat known material without changing the decision.
Synthesis
Move from source notes to claims, then from claims to structure.
Source summaries are not an outline. First extract candidate claims and connect each one to its support, limitation, and confidence. Then order the claims according to the reader's decision: context, recommendation, evidence, operating method, failure cases, and next action. This produces a narrative that can survive fact checking because the structure was built from claims rather than prose momentum.
The board should make dissent visible. If experience conflicts with product language or two tests produce different outcomes, keep the disagreement in the working model. A qualified conclusion is often more useful than a universal verdict, especially when integrations and interfaces change.
- Give consequential claims a source or mark them clearly as direct observation or inference.
- Keep a separate list of attractive statements that the available evidence does not support.
- Resolve or disclose contradictions before drafting the polished narrative.
Production handoff
Do not make the research board carry the entire media pipeline.
Poppy can hold the approved narrative, source relationships, and draft material, but recording, editing, design, and distribution need their own controls. An OBS capture should prove the workflow. CapCut can tighten pacing and captions. Canva can package reusable visual assets. The approved claims and sequence should travel with the handoff so convenience tools cannot quietly rewrite the conclusion.
For every downstream asset, identify the source of truth and the transformation allowed. A short may compress the argument, but it should not convert a conditional finding into an absolute promise. A newsletter may add context, but it should not detach a current claim from its date or evidence.
- Freeze an approved claim sheet before creating derivative assets.
- Record the complete demonstration before editing it into shorter forms.
- Review captions, generated voice, and summaries against the approved meaning—not only spelling.
Publish gate
The final question is not ‘does it read well?’
A publishable piece should answer four questions: what is the decision, what evidence supports it, where might it fail, and what should the reader do next? It should also identify affiliate relationships in context and avoid claims about price, availability, or behavior that were never verified for the publishing date.
The board remains useful after publication. Corrections, audience questions, changed product behavior, and new tests can be attached to the original reasoning. That creates an update path based on evidence instead of rewriting from memory when the article is already stale.
Continue the field guide