Neuronify

How we work

Corrections

We publish numbers, and numbers can be wrong. When ours are, we fix the page, say what changed, and keep a permanent record here. 2 entries so far.

The policy

What counts. A correction is for something we published that was wrong: a figure, a date, a name, a description of what someone did or said, or a method that produced a misleading number. A clarification is for something that was accurate and left out a fact a reader needed. Fixing a typo, repairing a link, refreshing daily data, or updating a guide because a law changed is an edit, and is not logged here.

What we do. We fix the page as soon as we have confirmed the error, and we aim to do that within two business days of hearing about it. The correction is logged below with the date, what was wrong, and what is right. Where the error affected a conclusion, the entry says so. Where it appeared in a newsletter issue, the next issue carries it too, because the people who read the mistake should read the fix.

What we do not do. We do not quietly rewrite a published number, and we do not delete an issue or an entry in this log. If a correction turns out to be wrong, it gets its own entry. We do not change historical data to make a series look smoother: when a method changes, the old rows stay as they were and are marked, and the methodology page records the change.

Who can ask. Anyone. If you work at a company we cover and we have a fact about it wrong, tell us and we will check it against a source and fix it. What we will not do is change an accurate figure because its subject dislikes it, or alter how a company is listed in exchange for anything. How we decide what to publish is in our editorial standards.

Report an error

Email support@neuronify.com with “Correction” in the subject, or use the contact page. Tell us the page, what is wrong, and, if you have it, a source for the right answer. A link to a primary document gets the fastest fix. We reply to every report, including the ones where we conclude the original was right.

The log

  1. · Correction

    NYC AI Hiring Index: tools named in job postings

    From August 17 to September 19, 2026 we counted a tool once for every posting whose full text named it. Several companies name a vendor, as a customer or an investor, in the introduction they repeat across all their job posts, so those vendors were counted once per open role. On the day we fixed it, OpenAI fell from 138 postings to 37 and Anthropic from 117 to 64. We now remove each company’s repeated text before reading a posting. The figure we publish and rank by, the number of companies naming a tool, moved by one company for four tools (OpenAI, Anthropic, Go, and Cohere) and not at all for the rest. No published finding depended on the posting counts. Data rows from before September 20 are marked as using the old method.

  2. · Clarification

    AI in NYC Show episode guide

    When this page was published on September 19, 2026 it described Neuronify as the show’s media partner and the show as independently produced by its hosts. That was incomplete. Rob May, who co-hosts the show, is a co-founder of Neuronify. The page now says so.