A clearer view of your documents
Find the facts.
Improve the document.
See which claims are false, inspect the evidence and get a practical opinion on the quality of your document.
Start free · PDF, Word, PowerPoint, images or text
Illustrative example · fictional data
“There were 84 operating facilities in 2024.”
Page 3 · Key figuresThe result, made useful
Three answers. One clear review.
Start with what needs your attention. Open the evidence when you want to look closer.
Illustrative example · fictional data
Your review, at a glance
Factual strength
18 supported and 1 false · 19 of 31 claims scored
19 of 31 claims resolved with evidenceThe other 12 need more evidence.Start here
There were 84 operating facilities in 2024.
The source reports 42 for the same period and scope.The report uses data from 2024.
SupportedPrivate investment increased by 18%.
Comparable data is needed to confirm this.One fact. Two versions.
2024 · same scope
Industry survey · 2024
“The 2024 survey recorded 42 operating facilities.”Passage from a fictional source
Why it is marked false
Both figures refer to the same scope and year. The document doubles the total recorded by the source.
Document quality
An editorial opinion on the text
Original heading
Key results
Suggested improvement
Available capacity limits growth
Let the heading express the section’s conclusion.
What works well
The sequence of context, data and conclusions makes the report easy to follow.
Writing quality is assessed separately from whether the facts are supported.
Show the working
A finding is only useful if you can inspect it.
Follow a claim from the document to the source. Decide what to correct with the evidence in front of you.
See an example reportNo supporting evidence? The claim stays unresolved. An absence of proof does not make it false.
“There were 84 operating facilities in 2024.”
The source records 42 facilities for the same period and scope. A second review confirms the contradiction.
Source passage
“The 2024 survey recorded 42 operating facilities.”Illustrative register · 2024
Document quality
Clear structure. Correct the figure, cite the original register and revisit the conclusion that depends on it.A simple workflow
From upload to a better draft.
Read the full methodSources and models can be wrong. The report helps guide your review.
- 01
Bring your document
Upload a file or paste text. Review the cost estimate and privacy options before starting.
- 02
Let Facticy review it
Claims are compared with retrieved sources. Proposed false claims receive a second check.
- 03
Make your next edit
Read the findings, check what is unresolved and use the quality review to improve your document.
Keep the analysis close to your document
Read. Ask. Improve.
Locate a claim on its page and ask what you need to understand before making your next edit.
Market overview
84 operating facilities1
The source reports 42
Page 4 / 12Ask your report
A conversation about the analysis
What should I review first?
Start with the number of facilities on page 4. The document states 84 and the cited source records 42 for the same scope and year.
Then review the claims that still lack sufficient evidence.
The claim and its sourceA closer reading
Writing it? Reviewing it? Start with the same evidence.
For reports you prepare, presentations you receive and decisions that deserve a closer look.
Explore document reviewYour document, your choices.
EU storage, encryption and retention controls. Analysis uses AI providers; confidential mode limits external research and still sends document text to those providers.
Read our privacy policyFrequently asked
Before you upload.
How does Facticy check a claim?
Facticy retrieves source passages and compares the claim’s subject, date, unit and scope. Another reviewer checks proposed false claims independently. If the evidence is insufficient or the comparison is inconclusive, the claim stays unresolved.
What can I upload?
PDF, Word (.docx), PowerPoint (.pptx), images (PNG, JPG or WebP), or pasted text. File and page limits depend on your plan. Image analysis is disabled in confidential mode.
Are my documents used to train AI?
The configured providers are selected for API data terms that exclude model training. Analysis still sends document text to AI providers, including in confidential mode. The privacy and subprocessor pages explain how that processing works.
Where are my files stored?
Files are stored in the EU, encrypted in transit and at rest. You can keep them until deletion, delete the original after analysis, or select automatic deletion after 30 or 90 days, subject to your plan.
What does unresolved mean?
There is not enough evidence for a clear conclusion. The reason may be a missing or inaccessible source, internal data, a forecast, disagreement or an incomplete check. Unresolved does not mean false.
Can I rely on the report alone?
Use it to guide your review. Sources and models can be wrong, and extraction or retrieval can miss relevant context. No detected false claims does not mean the whole document is accurate. Inspect the evidence and unresolved claims before making a decision.
Your next step
Give your next document a closer look.
Find what needs correcting. Understand what needs another source.