Every claim AI makes about you, checked against the truth.
Fact Check reads every AI answer that names your brand, pulls out each claim about your price, plans, free trial, integrations and company, and checks it against facts you confirmed. Wrong and outdated claims are listed with the quote, engine and prompt until a scan comes back clean.
A buyer asks ChatGPT what Kestrel costs.
20-person team, shortlisting three tools.It answers $12 per user, with a free plan.
Fluent, confident, wrong.Your pricing page says $40 and a 14-day trial.
The buyer feels misled.Tab closed. No demo, no reply.
Nothing in your CRM to tell you why.Fact Check finds the $12 before your next buyer does.
Example · fictional brand
The six things buyers ask AI before they ever visit your site. If any of them is wrong, the deal starts with a correction, or never starts.
Example claims for a fictional brand
Your own website is the source of truth. A model reads; MentionX's code decides.
Your pricing, plans, about and integrations pages. Only what they state.
Confirm, edit or reject each one once. Rejected facts are never used.
Each stored answer that names you, claim by claim, after every scan.
No fact or low confidence means unverified, never an accusation.
Tap every claim you think is wrong, then run the check. Kestrel's facts are on the right.
"Is Kestrel a good fit for a 20-person team?"Every wrong claim, with the evidence to act on it. In the app under Pulse → Fact Check.
Built never to accuse without evidence. Wrong and outdated need a confirmed fact and at least 60% confidence. Anything less is unverified.
A scan stores what every engine said about you.
Fact Check lists the claim with its quote, engine, prompt and your fact.
The Action Board ranks the fix; Creative Studio drafts the page engines should read.
The next clean scan moves the claim to Fixed, with its history.
Fact Check finds and proves the problem. It does not edit what an engine says; better sources do.
| Aspect | Asking yourself | MentionX Fact Check |
|---|---|---|
| Coverage | One engine, one phrasing, once | Every engine on your plan, every buyer prompt |
| Source of truth | What you remember | Facts from your site, confirmed by you |
| Evidence | A screenshot | Quote, engine and prompt, stored |
| Over time | Nothing recorded | A dot per scan and a Fixed state |
| Fairness | Your judgement | No fact, no verdict; 60% confidence minimum |
Still unsure? Talk to the founder.
Fact Check reads every AI answer that names your brand, pulls out each claim about your price, plans, free trial, integrations and company, and checks it against facts you confirmed. Wrong and outdated claims are listed with the quote, engine and prompt until a scan comes back clean.
From your own website first. MentionX reads your pricing, plans, about and integrations pages and writes down only what they state. You confirm, edit or reject each fact, and rejected facts are never used.
No. Fact Check finds and proves the wrong claim. The fix is better sources: the Action Board ranks what to change, Creative Studio drafts it, and the next scan shows whether the claim is gone.
After every scan, so weekly with weekly monitoring. A wrong price does not change day to day, so a weekly check catches it and shows when it is fixed.
Every engine on your plan. Fact Check reads the answers MentionX already stored for your scan, and no others.
A claim is only flagged wrong or outdated when there is a confirmed fact to check it against and the confidence is at least 60%. Otherwise it is unverified. Every claim carries its quote, so you can check it yourself.
Spark, Nova and Galaxy. Starter scans and measures only, so it does not include Fact Check. There is no extra charge on the plans that include it.
Run a scan and see every claim engines make about your brand, checked against your own facts.
