Now with parallel multi-model verification

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.

No credit card 5 free runs / month Export to Markdown
sourcescouter.com / verify / seo-simple-version.md

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.

Claims - 4
Verified3
Low conf.1
Sources17
Active pipeline
GoogleFirecrawlGPT-4o
Trusted by editorial teams at
The AtlanticStripe PressSubstackAhrefsBloombergFigma
How it works

Three steps from draft to cited.

Paste your article or upload a markdown file. SourceScouter handles the rest - extract, search, verify.

01

Extract claims

An LLM reads your draft and isolates every factual claim - leaving opinion, anecdote, and framing alone.

→ extracted 4 claims in 2m 3s
  ↓ #1 Google has gotten good at...
  ↓ #2 Internal linking helps users...
02

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.

GoogleFirecrawlBasic HTTP
03

Verify & annotate

Multiple models weigh each source's evidence, surface contradictions, and mark your text with a confidence score you can trust.

✓ Verified (84%) 3 sources confirmed
What's inside

Built for writers who cite their work.

Every feature exists to let you spot a weak claim before your reader does.

★ Flagship

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.

Reliability

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.

SERP
GoogleFirecrawl
Crawler
Basic HTTPFirecrawl
LLM
Gemini 2.5GPT-4o
Transparency

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.

developers.google.comFirecrawl
seo-theory.comHTTP
rebelmouse.comHTTP
Signal

Confidence that means something.

Five tiers - Verified, Low Confidence, Unverified, False, Pending - each backed by model agreement and source count, not vibes.

VerifiedFalseLow conf.Unverified
Workflow

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.

Why bother

The old way vs. a verified draft.

Before

Writing without a net

You half-remember a stat from a newsletter and type it confidently.
A reader emails to say you're wrong - you scramble through 40 open tabs.
Your editor asks for sources; you spend 2 hours rebuilding the research trail.
A contradiction sits in your post for 8 months before anyone notices.
After

Every claim earns its place

Paste your draft; every factual claim comes back highlighted and scored.
Low-confidence claims are flagged before publish - edit or back up with a primary source.
Sources live inline - hand them to your editor with one export.
Contradicting sources are surfaced with AI reasoning, not silently dropped.
Pricing

Pay for what you publish.

Most writers stay on Free forever. Upgrade when your draft schedule says so.

Free
$0/ forever

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
Start free
Most popular
Writer
$19/ month

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
Start 14-day trial
Team
$79/ month

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
Talk to us
FAQ

Questions writers ask.

Which LLMs do you use to verify claims?+
We run each claim through two models - typically Gemini 2.5 Flash and GPT-4o Mini - and only mark a claim "Verified" when they agree. You can see exactly which models voted on each claim in the inspector.
Does it work for non-English content?+
Today: English and Spanish are fully supported. French, German, and Portuguese are in beta - accuracy may be slightly lower on idiomatic claims.
How do you handle contradicting sources?+
We don't silently drop them. Contradicting pages appear in the URL list with a red flag and an AI Reasoning block explaining exactly why the page disagrees with the claim - so you can decide whether to revise.
Can I use my own API keys?+
On the Team plan, yes - bring your own OpenAI, Anthropic, or Google keys and we only bill you for the orchestration.
What happens to my drafts?+
Drafts are encrypted at rest, never used for model training, and deleted permanently within 30 days of a run unless you pin them.

Stop publishing things you can't back up.

Verify your first article in under three minutes. No card required - just paste and go.