VeraGen goes live December 1. New subscriptions open then. See the $5,000 Launch Challenge →

Home / Case Studies / Syrion.ai

Case study · Syrion.ai

The clear view gained through inquiry.

Syrion gives researchers and independent investigators AI-assisted access to the public record: declassified documents, patents and other open intelligence. Ask a question and get an answer written only from the documents, with a numbered citation after every claim. It is live at syrion.ai, built 100% with VeraGen. It is not affiliated with or endorsed by the CIA or any government agency.

Syrion.ai 100% built with VeraGen

An AI research platform for declassified and open records, live · Syrion.ai

Case Study: Syrion.ai on YouTube

Watch Syrion.ai being built.

Nothing staged: each episode is recorded in the real product as the company is built. 4 episodes so far, 10 minutes.

Episode 1 · Brand Studio · 3:05

A Brand You Can Trust with Declassified Records

The brand in Brand Studio: a brief that writes down the words to avoid, a vault-door logo with no agency imagery, the reading-room palette and an archivist’s voice, approved as v1.0.

The playlist

Every episode, in order.

Watch the full playlist on YouTube
Video thumbnail: A Brand You Can Trust with Declassified Records3:05

Episode 1 · Brand Studio

A Brand You Can Trust with Declassified Records

The brand in Brand Studio: a brief that writes down the words to avoid, a vault-door logo with no agency imagery, the reading-room palette and an archivist’s voice, approved as v1.0.

Video thumbnail: Collecting a Million Documents2:43

Episode 2 · Tellagent

Collecting a Million Documents

Tellagent data pipelines with no custom code: the 1,003,493-record catalogue, the National Security Archive’s CIA collection, the full text, cia.gov’s open pages, Vault 7 and the nightly update.

Video thumbnail: Search That Ranks and Cites1:59

Episode 3 · Data Management

Search That Ranks and Cites

The tables in Data Management, ranked full-text search with highlighted passages, the document reader, and an AI answer with a numbered source after every claim.

Video thumbnail: From Spec to Research Workspace2:37

Episode 4 · Chimp

From Spec to Research Workspace

The website built in Chimp: the interview, a spec of six epics and thirteen pages, the design, the build, QA at 100s, the release and the live research workspace.

More episodes are on the way. Follow the playlist on YouTube to see each one as it lands.

What Syrion.ai does

Ask. See. Source.

The CIA Electronic Reading Room holds 1,003,493 declassified documents. Syrion reads them from the Internet Archive’s complete public copy, because cia.gov blocks automated collection, together with the open pages of cia.gov. Every scan and its text is copied to Syrion’s own storage, so the archive does not depend on another site staying up; the National Archives’ JFK Assassination Records and WikiLeaks’ Vault 7 are being added the same way. The CIA records are public-domain U.S. government records.

AI answers, cited

Full-page AI search. Answers are written only from the documents, with a numbered citation after every claim.

Ranked full-text search

Search every document, with highlighted passages, filters by year and a timeline.

A reader for every document

The text, the scan (PDF) and the CIA document number, with links to cia.gov and the Internet Archive.

Research documents

Generate a brief, a timeline, a summary or an annotated bibliography, and download it as PDF, Word or Markdown.

Sources and sharing

Download every source as CSV, RIS or BibTeX, and share answers, documents and sources with links that unfurl as social cards.

Save your research

Optional free sign-in with a VeraGen ID to keep your research. Searching and reading need no account.

See it live at Syrion.ai

How it was built

100% VeraGen, layer by layer.

Here is which VeraGen app built each part of Syrion.ai, and the episode that shows it.

  1. Brand

    Syrion: a two-hook S mark, slate #0F172A and deep teal #0F766E, and the tagline “The clear view gained through inquiry.”

    Episode coming
  2. Data pipelines

    Tellagent collects new documents nightly, so the archive stays current.

    Episode coming
  3. Data & search

    The document tables in Data Management, and the new research search the answers are written from.

    Episode coming
  4. Website & app

    The website and the research app, built in Chimp.

    Episode coming

The results

What it adds up to.

1,003,493declassified documents in the archive, from the CIA Electronic Reading Room
4 appsBrand Studio, Tellagent, Data Management and Chimp
Nightlycollection of new documents by Tellagent
Every answercited: a numbered citation after every claim

More case studies

Built the same way.

All case studies

Build your company the way Syrion.ai was built.

Start with every VeraGen app in one account, or have our team build it with you on VeraGen, the same way these were made.