Claim verification
Hand over a generated statement about a company and get back the facts that support or contradict it, each with the dated capture it was read from.
Evidence states, not a confidence score. You see what was observed.
The job
The expensive errors are the plausible ones.
A model writes that a vendor is SOC 2 certified, prices from $99, and integrates with your warehouse. Every clause is the kind of thing that is usually true, which is exactly why nobody checks it, and why the one that is wrong reaches the customer.
Bixel answers the narrower question a record can actually answer: what was observed on the company's own surfaces, and when. A claim the record supports comes back with its source; a claim it cannot substantiate comes back saying so, which is the honest answer and the useful one.
What that costs today
- Review queues that scale with output volume
- Spot checks that sample the easy claims
- “Sounds right” as the last line of defense
- Corrections that arrive after the customer read it
The API
4 calls cover the job.
Current state is open. History values, claim checks, and evidence trails sit behind a free key. These are the real routes.
/v1 /companies /{company} /claimsCompany-stated claims lined up against what was observed.
/v1 /companies /{company} /factsThe observed side, one dimension at a time.
/v1 /companies /{company} /evidenceThe capture trail behind any fact, point-in-time.
/v1 /companies /{company} /facts /{key} /proofA proof bundle the reader can verify without trusting Bixel.
For agents · MCP
Or let your agent run it.
Connect the Bixel MCP server and any agent, Claude, an internal copilot, a procurement bot, works this job against live structured data, with every value citing the dated capture it came from. The recommendation becomes auditable, not a guess.
You ask your agent
“Here is a paragraph we drafted about Pinecone. Which sentences does the record support, and which cannot be substantiated?”
Check the sentence before it ships.
Pinecone's security page separates what the company states from what was independently observed: see claimed against observed →
More in AI & Agents
The answer layer for agents that need company facts with receipts.
Agent grounding over MCP
Connect the MCP server and your agent answers company questions with dated, sourced facts.
Explore
RAG corpora
Licensed slices of the record, by category or dimension, as retrieval sets for vertical AI.
Explore
Factuality benchmarks
Dated ground truth for evaluating models on what was true as of a given date.
Explore
Entity resolution
Domain to company to parent, with subsidiaries and renames handled.
Explore
Procurement agents
Agents that shortlist, compare, and monitor vendors use the record as their memory.
Explore