We build research integrity software and verification workflows.

Our manifesto

Because accuracy matters.

Evidite’s verification engine uses no artificial intelligence or machine learning.

Fully deterministic
Every result traceable
Independently reproducible

Built for every professional who cites sources

A lab researcher in a hairnet and mask working at a laptop beside a microscope

Folio

for publishers, journal editors and research integrity offices

Checks every reference in a published article against the record. Folio flags every DOI that is not registered or opens a different work, and proposes a DOI wherever it can identify the cited work. Evidite runs it as a service for whole collections of articles.

Learn more

When AI hallucinations hit the courtroom

Bloomberg
Sullivan & Cromwell Apologizes to Judge for AI Hallucinations
April 21, 2026
The New York Times
A.I. ‘Hallucinations’ Created Errors in Court Filing, Top Law Firm Says
April 21, 2026
CNN
Another ‘hallucinated’ court filing highlights the difference between Silicon Valley and the rest of the world
April 23, 2026

AI is making citation errors worse — not better

The headlines are writing themselves. The crisis is real, accelerating, and unsolved.

Hallucinated citations are polluting the scientific literature
An estimated 110,000+ publications with invalid references in 2025
View source
US Appeals Court Fines Lawyers $30K Over AI Filing Errors
Sanctions escalating 6x in 3 years
View source
Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools
Lexis+ AI & Westlaw AI: 17–33% hallucination rate
View source

Enterprise-grade infrastructure

29 Data Sources
CrossRef, PubMed, OpenAlex, Google Books, Library of Congress, CourtListener, and more.
Verification Engine
Confidence scoring and retraction detection.
Word Integration
Native Microsoft Word add-in. Resolve inline or batch-process entire documents.
Patent-Pending IP
One provisional utility patent filed. Two utility patents in progress after prior art searches found no blocking references.
Built to scale
Cloud-native on Google Cloud Run. Designed for 10,000+ simultaneous users.
Cost-Optimized
Waterfall architecture minimizes expensive API calls. Cache-first strategy cuts costs 90–99%.

Made by a different kind of company

Built by a writer, for writers, as well as researchers, editors, and lawyers. Good tools come from understanding real people and real needs. We’re an innovative team led by collaborative research and design thinking.

Learn about our team