SecondSourceThe Second Source on AI

About SecondSource

SecondSource is an AI-industry research publication. It uses AI to read primary documents across fields — filings, papers, regulations and what industry leaders say on the record — attaches sources and verification status to its judgments, labels what has no source as unverified, and corrects mistakes in public. Published daily in English, Traditional Chinese and Japanese.

Who's behind it

Founded on July 8, 2026 and based in Chicago, SecondSource is self-funded by its founder, who serves as editor-in-chief and is responsible for its rules, key editorial calls and content. AI does the reading, drafting and checking under written rules, the main ones published on the methodology page. Contact: hello@secondsource.io

To cite or republish SecondSource, see Press and citation.

What AI does here

AI reads the sources, drafts, and writes the brief in Traditional Chinese, English and Japanese. Facts in the body link to the original documents as a rule, the writing rules require that only what the original supports goes in, and news we have not verified is labeled as such; when an English-language post on X is quoted verbatim, a program also checks the quote word for word against an archived copy of the post. Before publication, a separate AI review process checks the draft against written rules, and a draft that fails is not published; not every piece is read word for word by a person. The editor-in-chief is responsible for the rules and corrections; daily checks and weekly sample reads of published issues are run by an AI acting as editor, under those rules; mistakes go on the corrections page, and entries there are never taken down.

Why there's no byline

Articles carry no author byline; the publication answers to its editor-in-chief. The credibility of each piece comes from verifiable evidence, not titles: judgments link to their sources, and forward calls also get a public verdict card; only published information is used, paid publications included, and when a paywalled source is cited the page says how much of it was read; corrections are made in the open.

The brief and the research graph

The brief is useful when it helps readers see how technology turns into products and changes companies and industries; an item doesn't have to lead to an immediate decision. Insights that go into the research graph (the database where facts and judgments accumulate) clear a higher bar; see Methodology.

Where to read it

  • Website: the full text of every issue, with sources and verification notes.
  • Email: the Chinese and English editions go out as a shorter email that links back to the full text on the site.
  • Japanese edition: published on note, with the full text and verification on the Japanese site.

Principles

  • Judgment over volume. The goal isn't more information; it's more accurate judgment.
  • Not a news digest. AI reads primary sources across fields; the focus is what has changed and which claims still lack evidence, not what happened today.
  • Sources attached, mistakes corrected in public. Judgments carry their source and verification status, what has no source is labeled unverified, and mistakes are logged on the corrections page (details on the Methodology page).

Subscribe

Calls in the daily brief on enterprise procurement, platform selection and information security from the last 7 days are gathered on the procurement and security calls page.