Turn collected research into a provenance-aware brief
Organize a target query, research questions, entities, keyword questions, and supplied competitor topics without implying that a model or live search engine discovered them.
AI Content Brief Generator
Free, no signup. Compose a brief from your research inputs — and, optionally, the competitor pages you enter. Every section states whether it was observed, heuristic, or not evaluated—an unavailable engine never becomes a silent gap.
Competitor pages to research (optional)
Enter up to five pages that already rank for this topic. We fetch only the URLs you enter — no search results are scraped — and read each page's headings, length, entities, and questions as observed evidence. Paste a page's content to skip fetching it.
Checks run from our server; we fetch the URL you enter and don't keep the results. Pasted content and the assembled brief stay in your browser. A competitor URL is sent to the site's bounded fetcher only when its paste field is empty; no search results are scraped. Anonymous run-level outcome counters may be used for aggregate research; URLs, domains, IPs, and identifiers are never included, and no statistic is released below 100 runs.
Site passport Local context for this saved site
Local data
Saved targets, named lists, and recent check summaries remain only in this browser.
Rate this tool
How to use it
- Enter the target query.
- Paste only the questions, entities, keyword ideas, and competitor topics you have actually collected.
- Optionally enter up to five competitor page URLs (or paste their content) to record their observed outlines, common headings, entities, and questions.
- Select Assemble brief and review the state printed beneath every section.
- 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:
Research questions / fan-outobserved2 supplied questions
Entity checklistheuristicGoogle Search Console, robots.txt, XML sitemap—all unlinked
Questions to answerheuristic2 supplied keyword questions
Competitor topics supplied by youobservedcrawl prioritization; validation after fixes
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
Does the AI Content Brief Generator scrape search results or call an AI model?
No. It never queries a search engine or an answer engineAnswer Engine Optimization (AEO) is the practice of structuring content so engines deliver it as a direct answer — featured snippets, voice assistants, and AI search — rather than just a ranked link. Coined for voice search in 2018 and revived for the LLM era. Google's position is that it's still SEO., never scrapes a results page, and no model writes the brief. If you enter competitor page URLs, it fetches those specific pages through the site’s bounded fetcher (paste-first) and records what it observes — headings, length, entities, and questions — as observed evidence. Empty sections stay explicitly not evaluated.
What do observed, heuristic, and not evaluated mean?
Observed marks user-supplied research questions and competitor topics, plus anything read from a competitor page the tool fetched. Entity names"Entity & identity schema" is a practitioner label — not an official Google or schema.org category, and not exhaustive of identity-capable schema.org types — for the three main types that declare who or what is behind a site: Organization, LocalBusiness, and Person. and keyword questions entered here are marked heuristic. Empty source sections remain explicitly not evaluated.
Why are supplied competitor topics labeled observed?
Observed means a string was supplied as source evidence to the composer, or (for fetched competitor pages) read from a page the tool retrieved. Supplied topic strings are not independently verified against a live SERP.
How does the optional competitor fetch work?
Enter up to five page URLs, or paste their content. A URL is fetched only when its paste field is empty, through a bounded fetcher capped at 2 MiB per page and rate-limited. The tool reads a conservative main-content approximation for headings, word count, entities, and question headings. A page that cannot be fetched is reported as not evaluated, never faked.
What can I export?
You can copy a Markdown brief or a CSV with section, state, and item columns. Both retain provenance states.
Is my brief uploaded or saved?
No. Composition and export happen locally in your browser; only the competitor URLs you enter are fetched, and only when their paste field is empty.
Feature requests for Ai Content Brief Generator
Upvote what you want most. New ideas can be submitted from the floating Feedback menu; requests appear here once approved, and the most-wanted rise to the top.
You won't be emailed about that request anymore.
Loading…
➕ Request a feature
New requests are reviewed before they appear here.
Where this tool helps
Common use cases
Compare two competitor outlines without scraping a SERP
Paste page content or provide up to five specific URLs to extract observed headings, word counts, entities, and question headings from only those pages.
Keep research gaps visible
Preserve empty or failed source sections as not evaluated instead of silently dropping them or manufacturing completeness.
Create an editable editorial handoff
Copy a Markdown brief that keeps every section's observed, heuristic, or not-evaluated state attached to the planning evidence.
Export a structured QA matrix
Copy CSV rows with section, state, and item columns for spreadsheet review, assignments, and follow-up research.
Watch the full workflow
AI Content Brief Generator walkthrough
Read the transcript
AI Content Brief Generator
This beginner walkthrough turns a fictional website-migration research packet and two pasted competitor pages into an organized content brief. We will first show how missing research stays visible, then build the fuller brief, explain every evidence label, inspect page outlines and questions, and copy the Markdown and C-S-V handoffs. No A-I model or search-results scraper runs.
Step 1
The name can be confusing, so begin with the boundary: this page does not ask a language model to invent a brief. It organizes the research you type or paste and the exact competitor pages you choose. Every section says whether it came from supplied evidence, a simple planning rule, or no evidence at all. That makes it easier to see what you know and what still needs research.
Step 2
Use it to organize notes before assigning an article, compare the outlines of a few pages you selected, reveal missing research before drafting, give a writer an editable Markdown brief, or send a spreadsheet-friendly C-S-V to editors and subject-matter experts. It helps structure work you already collected; it does not replace the research itself.
Step 3
Start with the target query: the topic the page should answer. Fan-out questions are research questions you collected. Entity names are important people, products, organizations, or concepts. Keyword questions are search phrases you want the draft to address. The tool does not discover any of these for you, so enter only evidence you actually have.
Step 4
If your research notes already contain topics covered by competing pages, add one topic per line here. The brief calls these observed because you supplied the strings as evidence. That label does not mean the tool checked a live search result or proved that the topic is important. Record where your notes came from outside this brief.
Step 5
You can also research up to five specific pages. For each page, enter its U-R-L, paste its content, or do both. If pasted content exists, the tool uses it and skips the network request. Otherwise it tries the exact U-R-L through a bounded fetcher. From that page it records approximate word count, H-two and H-three headings, possible entities, and question headings. It never scrapes a search-results page.
Step 6
First, type only the fictional target query and select Assemble brief. The page creates a preview even though every supporting list is empty. That is useful for a beginner: an empty source is not the same thing as research that found nothing. The brief should expose the missing work instead of silently looking complete.
Step 7
The preview keeps all four base sections. Each contains a not-evaluated message because we supplied no fan-out questions, entities, keyword questions, or competitor topics. The page does not hide the empty sections or make up suggestions. Treat every not-evaluated line as a research task to complete before you call the brief finished.
Step 8
The complete fixture targets an enterprise website migration checklist. Two research questions, three entity names, two keyword questions, and two supplied topics form a bounded, inspectable packet. Nothing in this scene claims those strings came from a model or search platform.
Step 9
Two reserved example-dot-test U-R-Ls identify the evidence, while each adjacent paste box contains the fictional page H-T-M-L. Paste-first behavior means this walkthrough makes no page-fetch request and cannot accidentally depend on a changing external page.
Step 10
Now select Assemble brief with the full fictional packet. The status says the brief was assembled locally and that two of two competitor pages were researched. Because both pages were pasted, researched means their supplied content was read by the local extractor. It does not mean the tool visited or verified the live pages.
Step 11
The status confirms local assembly and the two-page count. Copy Markdown supports an editable writing handoff; Copy C-S-V produces section, state, and item columns for structured review. Neither export removes provenance.
Step 12
The research questions say observed because we supplied those exact strings. Entity names and keyword questions say heuristic, meaning the tool changed our inputs into planning prompts using simple rules without checking an outside source. Competitor topics are observed as supplied notes, not verified search findings. Always read the item and its label together.
Step 13
The competing-page section keeps each fictional U-R-L, an approximate word count, and its H-two and H-three headings. These are observations from the content we pasted. Use the outlines to notice subjects and ordering, then return to authoritative sources for the facts. A simple extractor can miss headings hidden behind scripts, navigation, or unusual page structure.
Step 14
Pre-launch checks and post-launch validation appear on both fictional pages, so the brief lists them as common headings. Repetition is a clue that readers may need those subjects. It does not prove the wording is correct or that your page needs the same section. Research the underlying need and write an original structure.
Step 15
The entity section counts names extracted across the pasted pages, including Engineering and S-E-O on both. This lightweight local extraction is useful for review prompts, but it is not semantic entity reconciliation and can include capitalized phrases that need human cleanup.
Step 16
The final observed section preserves two question headings: how teams should triage migration errors and what belongs in a first-day report. They are coverage clues from the supplied pages, not search-volume evidence or instructions to reproduce competitor wording.
Step 17
Select Copy Markdown, then Copy C-S-V. This capture reads both clipboard results and verifies that the evidence labels are present. Markdown is convenient for a writer or editor. The C-S-V repeats the state on each item, making observed evidence, rule-based suggestions, and research gaps easy to filter and assign in a spreadsheet.
Step 18
Observed means the string was supplied by you or read from a page you chose. Heuristic means the tool turned supplied text into a planning item using a simple rule without outside verification. Not evaluated means the required source was missing. Unlinked means an entity name was not connected to a stable identifier. None of these labels means a fact is automatically correct.
Step 19
The composer always creates four base sections, preserves empty inputs, and adds observed competitor sections only when pages are supplied. With two or more successful pages, normalized headings recurring on at least two pages get a separate common-heading section.
Step 20
Supporting features include state labels on every section, honest unavailable inputs, optional paste-first competitor extraction, a readable Markdown preview, and Markdown plus C-S-V copy actions. Browser-only composition keeps pasted content and the finished brief local.
Step 21
This tool does not generate prose, scrape search results, score content, estimate search volume, or reconcile entities. It can fetch only five U-R-Ls you provide, up to two mebibytes each, and simple extraction can miss rendered content. Pasted and fetched strings are not independently verified. Use the brief to organize research, then add source review, subject-matter expertise, editorial decisions, and original writing.
Use the brief as a research map—not a finished article.
Keep each evidence label attached when you hand the brief to a writer. Check the supplied and pasted observations, research the suggested entities and questions, and fill every not-evaluated gap. Use competitor headings only to spot topics worth researching, never as wording to copy. This brief is a planning map, not a finished article or an A-I verdict.
Common issues & how to fix them
- Warning Brief contains an uncovered validation topic Fix: Add a dedicated brief section or question that directly covers the flagged topic and defines what evidence the writer must include.
- Warning Brief lacks an evidence/source plan Fix: List the primary source, first-party data, expert input, or reproducible test needed for each important claim before drafting.
- Warning Brief does not cover a required intent Fix: Add a section whose format and answer satisfy the omitted intent—informational, comparison, transactional, or navigational—without diluting the primary query.
- Information Brief sections overlap Fix: Merge sections that answer the same intent, keep the stronger heading, and move any unique evidence requirements into that single outline item.