bearing

AI Disclosure

Last updated · June 14, 2026

The Heading is the one place Bearing uses AI. Think of it as an analyst that reads your own numbers and writes you a short, plain brief: what is working, what is not, and the one thing worth fixing next. Here is the short version of how it treats your data, with the full detail below for anyone who wants it.

The short version

  • Your customers stay anonymous. The AI never sees their names, emails, or phone numbers, and never sees a single individual at all. It only ever reads your totals, like how many people visited and how many booked.
  • It’s built to keep you on the right side of privacy rules. Bearing sets no cookies and needs no consent banner, and the AI works from grouped numbers, not anyone’s personal details.
  • It is there to help you get more customers. The Heading reads your own numbers and tells you, in plain words, the move most likely to bring in more business. You decide what to act on.

That is the short version. The rest of this page is the detail, for anyone who wants to see exactly what goes to the AI and what never does. The Heading runs only on paid plans, only for sites you own, and only when the brief is generated or you ask for a refresh.

What the AI sees

When the Heading generates, Bearing sends the provider:

  • The aggregate analytics for the site: the same figures shown on your dashboard, including visitor counts, conversion rate, bounce, average time on page and scroll depth, funnel completion, value driven, and top sources. These are grouped totals and averages, never individual rows.
  • Your own description of what the site is for: the intent text you write.
  • Your site’s own public pages: their page content (via fetch) and screenshots in mobile and desktop sizes, so the brief can critique copy and layout. These are the same pages anyone can visit. The fetch identifies itself as Bearing’s bot and is never counted as analytics traffic.
  • Public web research: searches about your business and its market, such as reviews, competitors, and category.
  • If you have connected Google Search Console: the aggregate search queries and landing pages already imported for your site, with their impression, click, and position counts.

What it never sees

This is the heart of it. The Heading is built on aggregate analytics, so there is no individual visitor data for it to see, and in most cases Bearing never collected any to begin with.

  • No IP addresses. Bearing never stores them. They exist only momentarily at the edge to derive a coarse location, then are discarded.
  • No individual visitor records, sessions, journeys, or profiles. The AI receives aggregate totals, not per-person rows.
  • No raw personal data in labels. Email addresses and long number sequences such as phone, account, or order numbers are redacted in the visitor’s browser before anything reaches Bearing, so that personal data is not in Bearing’s data to begin with.
  • No cookies, no cross-site identifiers, and no fingerprint that survives a day.

The closest thing to visitor-derived input is the aggregate search queries and the aggregate call-to-action and engagement labels. Those are grouped counts, already stripped of personal data and never tied to an individual.

How the provider handles your data

  • Anthropic processes these inputs on Bearing’s behalf to generate your brief, and does not use them to train its models.
  • Separate processors render the page screenshots and measure page performance.

Guardrails

The Heading is held to a tighter standard than a chatbot. Four guardrails keep it accountable:

  • Numbers stay deterministic. The AI is handed the measured analytics and is forbidden from inventing a statistic. A post-check drops any line whose figures don’t trace to a real measurement, so every number you read is one Bearing actually counted.
  • It shows its work. Every brief lists the exact searches it ran and the pages it read.
  • It never changes anything on its own. Goal-value and primary-conversion suggestions are suggestions; nothing is applied until you click Apply.
  • It is scoped tightly. Paid plans only, owned sites only, and only on generation or refresh.

More on data handling

This page covers the AI feature specifically. For how Bearing handles data across the whole product (analytics, accounts, retention, and your rights), see the Privacy Policy.