This article was AI-generated as part of an experimental historical-content project. The date reflects the period being analyzed rather than the date the article was originally written.
Last Thursday Google expanded Personal Intelligence to AI Mode. Google AI Pro and Ultra subscribers in the U.S. can now opt in to connect Gmail and Google Photos, so that AI Mode answers draw on their own messages and pictures along with the web. It is a Labs feature, in English, for personal accounts only. Google says it is not available to Workspace business, enterprise or education users.
Most of the commentary will be about privacy, which is fair. I want to ask a narrower question, one that matters to anyone responsible for how a company is described. When Search can read a customer’s inbox, which version of your company does it find there?
The inbox is an archive companies wrote
Think about what sits in a typical personal Gmail account. Order confirmations. Shipping notices. Renewal reminders. Promotional emails with prices and terms from two years ago. A support thread that ended badly. A newsletter announcing a product line that was later discontinued.
Much of that material was written by companies. Until now it stayed where it landed. With Personal Intelligence it becomes context for answers. Google’s own launch examples make the point. Robby Stein describes AI Mode noticing a sneaker brand he had just bought and suggesting a new style, and using a flight confirmation in Gmail to work out the destination and timing when recommending a coat. Google says AI Mode “considers the types of items you buy and where you shop.”
That is useful when the inbox is accurate and current. It is less useful when the most specific information the system can find about you, for that user, is an old promotion, a superseded return policy or a complaint thread.
The answer nobody else can see
Classic reputation work assumes a shared results page. A team can search a name, take a screenshot and assume most people see roughly the same thing. Personalization has always blurred that a little. This goes further. Two customers asking AI Mode about the same company could get answers built partly from material that neither the company nor any monitoring tool can see.
Google is candid about the limits. Its post says the system “might incorrectly make connections between unrelated topics or not fully understand the context,” and invites users to correct it with a follow-up or a thumbs down. That correction happens inside one person’s session. It does not reach the public sources that shape everyone else’s answers, and it does not reach you.
In 2023 I wrote that Bard’s “Google it” button admitted that answers still need corroboration. A personalized answer is harder to corroborate by design, because part of the evidence is private.
Where the work moves
What can a company actually do? A few things follow, none of them dramatic.
Treat customer email as public-facing content with a long shelf life. Transactional and marketing messages should carry clear dates, and terms that expire should say so in the message itself.
When a policy, price or product changes, tell existing customers by email, not only on the website. The inbox now competes with the website as a source, and the newest message on a topic is the one most likely to be read in context.
Keep the current version easy to find on the open web. A personalized answer still mixes the user’s context with what Search retrieves publicly, so a clear public record gives the private material something to be checked against.
And accept a new blind spot. Monitoring tools can sample public answers. They cannot sample answers built from a stranger’s mailbox. Customer-service logs may become the earliest signal that people are being told something odd.
The feature is small today: an opt-in experiment for paying U.S. subscribers. But on Wednesday Google said it will bring Personal Intelligence to Chrome in the coming months. Part of a company’s reputation is going to be assembled from things it sent, one inbox at a time.