AI Content Brief Generator

How to use it

1. Enter the target query.
2. Paste only the questions, entities, keyword ideas, and competitor topics you have actually collected.
3. Optionally enter up to five competitor page URLs (or paste their content) to record their observed outlines, common headings, entities, and questions.
4. Select Assemble brief and review the state printed beneath every section.
5. Copy Markdown for editing or CSV for a structured handoff. Research any not-evaluated section before treating the brief as complete.

Example output Example data — engine output

The interactive technical-audit example produces four sections:

Enter competitor page URLs above to add observed sections — competing outlines, common headings, shared entities, and questions — to the same brief.

What the results mean

Observed means the relevant input list was supplied to the composer. Heuristic means the page converted supplied entity or keyword strings into a planning checklist without external reconciliation. Not evaluated remains a visible placeholder when a source list is empty. “Unlinked” means an entity name was not reconciled to an identifier.

How it works

The composer trims the target query and creates a fixed four-section brief. Fan-out and competitor lists are passed through as supplied evidence. Entity names become heuristic, unlinked checklist items. Keyword ideas entered on this page are typed as heuristic questions. Empty arrays are never dropped: they become a not-evaluated item in Markdown and CSV. When you supply competitor page URLs, each is fetched (paste-first) and read locally for its H2/H3 outline, word count, entities, and question headings; those appear as additional observed sections, and headings that recur across two or more pages are called out separately.

Features

• Provenance state on every section.
• Unavailable inputs remain visible instead of becoming fake completeness.
• Optional, paste-first competitor fetch that records observed outlines, common headings, shared entities, and questions — no SERP scraping.
• Markdown preview plus Markdown and CSV copy actions.
• Browser-only assembly with an honest interactive example.

Limitations

• No model generation, SERP scraping, content scoring, volume, or entity reconciliation.
• The competitor fetch is bounded — up to five user-entered URLs, 2 MiB each, rate-limited — and reads a conservative main-content approximation, so extracted outlines can miss rendered or off-canvas headings.
• Extracted competitor headings are coverage inspiration, never phrasing to copy.
• User-supplied and fetched strings are not independently verified.
• The fixed four-section base plus optional competitor sections is a research handoff, not a complete editorial specification.
• The page labels all entered keyword questions as question intent regardless of phrasing.

Frequently asked questions