Research & editorial review
Fact-checking for reports that get relied on
Analysts, researchers and publishers whose reports get quoted by other people.
The number that survived the rewrite
A market report is updated with fresh data. One figure from last quarter is left untouched, and it now contradicts the table two pages later. The report is published. Everyone who quotes it inherits the mistake, and the correction never travels as far as the original did.
in/guard/out re-checks a document every time it changes, not just on the first draft. It redoes the maths, matches each figure to its source, and marks the claims stated more confidently than the evidence allows.
Research doesn’t fail at publication – it fails at revision. A report goes out checked; then new data lands, sections get updated, and a stale figure or a citation that no longer holds slips into version three. Whoever cites it inherits the error.
in/guard/out fact-checks the way an editor would, but on every pass. Numeric grounding recomputes the arithmetic and matches every figure in the text to its source. A fact-check judge flags claims the evidence cannot support, contradictions introduced by an edit, and overconfident extrapolation – while explicitly rewarding an honest “I don’t know” over a confident guess.
Claims that need outside verification are cross-checked against real sources: your own document store via RAG, live web search, or a second model acting as a critic. And because every revision runs through the same wire, re-checking an updated report is not a special workflow – it is just the next request.
Where it breaks
- ✕Fabricated citations & invented references
- ✕Stale figures carried into an updated report
- ✕Claims overstated beyond the evidence
- ✕Contradictions introduced when a report is revised
What answers it
- →Fact-check judge – flags overconfident & false claims, rewards “I don’t know”
- →Numeric grounding – recomputes the arithmetic and matches every figure to its source
- →RAG, web search & consilium cross-check against real sources
- →Re-checked on every update – each revision re-verified against current sources
- →Verify claims against private data via MCP – e.g. a market-share figure only your own data can confirm Coming soon
The stale market-share figure
A quarterly report is updated with fresh numbers, but one market-share figure survives from the previous version and now contradicts the new table two pages later. The grounding pass flags the figure as unsupported by the updated source; the fact-check judge flags the internal contradiction. Both findings land in the request’s grounding report with the exact claims quoted.
Frequently asked questions
Can it re-check a report after we update it?
Yes – every revision that passes through in/guard/out is verified against its current sources, the same as the first draft. Re-checking on update is the normal path, not a special one.
How does it handle uncertain claims?
The fact-check judge is instructed to reward honesty: a hedged estimate or an explicit “I don’t know” isn’t flagged. Only unhedged certainty the source cannot support is.
What sources can claims be checked against?
The material in the request itself, your own vector store via RAG, live web search, and a second-model consilium. Verification against private data over MCP – for claims only your own data can confirm – is coming soon.
Does it catch made-up references?
Citations and figures that appear nowhere in the provided sources are flagged as unsupported by the grounding pass and the judge; in prevent mode the response is blocked before it ships.
Other clients
See it on your own traffic.
We’re running a limited demo – sign up and we’ll get you in as soon as we can.