# Make the claim answer to the evidence.

An AI claim-and-citation checker shipped 0→1. Document review went from weeks to hours, with 1,000+ issues flagged.

Company: VectorShift

Contribution: 0→1 product delivery

Company tenure: VectorShift · Jul 2025–present. This is not a claim about individual project dates.

Context: 1,000+ issues flagged

## TL;DR

Problem: Checking whether every claim in a document was backed by its citation took reviewers weeks.

My call: Check each claim and its citation as a pair, and give reviewers a list of issues instead of a document to reread.

Outcome: Review went from weeks to hours. 1,000+ issues flagged.

- Review workflow: Weeks → hours
- Issues flagged: 1,000+
- Shipped from scratch: 0 → 1

## Context

- Company: VectorShift (YC S23)
- My role: Shipped it from zero to one
- Users: Teams reviewing financial and diligence documents
- Framing: Quality as a product problem, not a QA step

Conceptual review flow. Claims and their citations feed an issue-review workflow. Flagged issues are not a guarantee of completeness.

## The problem

Financial and diligence documents make claims and cite sources for them. Someone has to check that the two actually match.

A citation being there says nothing about whether it supports the claim. A number gets updated in one place and not another. A source gets paraphrased until it says something different.

Reviewers were doing this by hand, and a review took weeks.

Evidence: Review workflows took weeks before Verification. Most of that time went into confirming claims that were fine, in order to find the few that were not.

## Constraints & non-goals

The product has to judge the relationship between a statement and its evidence, not just whether a citation exists.

False confidence is worse than no tool. The product flags. A reviewer decides.

Every flag has to show the claim and the cited passage side by side.

Out of scope: Certifying a document as correct. Rewriting the document automatically. Fact-checking against the open web.

## Options I weighed

Rejected: A citation-presence checker Easy to build and close to useless. The costly errors are cited claims that the source does not support.

Rejected: An AI summary for reviewers A summary is one more text to trust. Reviewers needed fewer things to read, not a different thing.

Chosen: Claim-and-citation checking with an issue queue It targets the real failure and turns review from rereading into triage.

## The spec

Users: Analysts who review documents before they go out. Leads who sign off on them.

Requirements: Pull claims and their citations out of a document. Check whether each citation supports its claim. Flag issues for review. Show the claim, the citation and the reason for the flag together.

Acceptance criteria: Every flag links to the exact claim and cited passage. A reviewer can dismiss a flag in one action, and the dismissal is recorded. A document with no flags still shows what was checked.

Guardrails: The product never labels a document “verified”. It reports the issues it found. Dismissal rate is tracked, so noisy checks get fixed instead of ignored.

## What shipped

1. Extract. Claims and their citations are pulled from the document.

2. Pair. Each claim is matched to the passage it cites.

3. Check. AI checks whether the evidence supports the claim.

4. Triage. Reviewers work through flagged issues instead of rereading everything.

## What changed

Review workflows went from weeks to hours.

Verification has flagged 1,000+ issues.

Quality became a surface reviewers use every day, not a checklist at the end.

Issues flagged is not the same as errors prevented, and no flag list is guaranteed complete. Both numbers describe the workflow, not accuracy.

## Reflection

Quality is a product problem, not a QA process. Treating it that way changed what we built.

If I did it again, I would instrument dismissals from the first release. The ratio of dismissed to accepted flags is the best single signal of whether reviewers trust the tool.

The takeaway: Check the claim and its citation together.

Last updated: 2026-09-13