Find sources for every claim in your content.
SourceScouter reads your article, extracts every factual claim, and hunts down primary sources across the web - annotating your text with confidence scores and evidence you can cite.
You don't need 50 tactics. You need a few things done consistently right. Most sites fail because they half-do everything.
Every time you publish, ask: where does this belong in my existing content graph?3 Then link to it from 3-5 related pages. Internal linking helps users stay longer.2
It also helps Google understand your site structure3 - and as a result, more content leads to more internal links, which leads to more authority, and then to easier rankings.4
Google has gotten good at sniffing out fluff.1 Every paragraph should earn its place.
Three steps from draft to cited.
Paste your article or upload a markdown file. SourceScouter handles the rest - extract, search, verify.
Extract claims
An LLM reads your draft and isolates every factual claim - leaving opinion, anecdote, and framing alone.
↓ #1 Google has gotten good at...
↓ #2 Internal linking helps users...
Search & scrape
For each claim we generate search queries, pull the top URLs, and scrape the actual page - with automatic fallbacks when a service is down.
Verify & annotate
Multiple models weigh each source's evidence, surface contradictions, and mark your text with a confidence score you can trust.
Built for writers who cite their work.
Every feature exists to let you spot a weak claim before your reader does.
Inline annotation, not a PDF report.
Claims are highlighted in your original prose with color-coded confidence. Click any highlight to see the source quote, the search queries that found it, and the models that weighed in.
Export the annotated version as Markdown with linked citations - ready to drop into your CMS.
Every time you publish, ask where it belongs. Internal linking helps users stay longer.2
It also helps Google understand your site structure.3 But more content = better rankings4 is overstated.
Pipelines that don't quit.
When Firecrawl is down, we fall back to Google Search, then raw HTTP. When one LLM rate-limits, another picks up the claim.
See every URL we considered.
No black-box confidence score. Every claim lists the pages we read, the queries used, and why a source was rejected.
Confidence that means something.
Five tiers - Verified, Low Confidence, Unverified, False, Pending - each backed by model agreement and source count, not vibes.
Export as Markdown with citations.
One click produces an annotated .md file with inline [^source] references. Drops into Ghost, Substack, Notion, or any static-site generator.
The old way vs. a verified draft.
Writing without a net
Every claim earns its place
Pay for what you publish.
Most writers stay on Free forever. Upgrade when your draft schedule says so.
For personal essays, newsletters, and learning the craft.
- ✓5 verified articles / month
- ✓Up to 25 claims per article
- ✓Markdown export
- ✓History of last 20 runs
For working writers who publish weekly and want the full source trail.
- ✓Unlimited articles
- ✓Unlimited claims per article
- ✓All export formats (MD, DOCX, PDF)
- ✓Full pipeline transparency
- ✓Priority LLM queue
For editorial teams with shared style, shared sources, and shared standards.
- ✓Everything in Writer
- ✓Up to 5 editors
- ✓Shared source library
- ✓Claim-level comment threads
- ✓SSO & audit log
Questions writers ask.
Stop publishing things you can't back up.
Verify your first article in under three minutes. No card required - just paste and go.