Portraits

What Portraits is

Portraits is a Google Labs experiment that lets you hold a conversation with an AI representation of a trusted expert, beginning with well-known authors and coaches. The address listed in the catalogue now returns a not-found page, and the experiment appears to have moved to a separate domain, so this entry reflects a thin source and a product in motion.

What it does

You open the experiment and talk with an AI version of a named expert that draws on their published thinking and voice. The loop is ask, listen, reflect, aimed at people seeking guidance on work and career questions.

Public material describes it as an experiment launched in 2025 with participation from named authors whose expertise and voice are represented. Because the original Labs path now returns an error while another domain presents the same idea, the practical first step for anyone interested is to find the current home of the project rather than trusting the catalogue link. As a research experiment it may change, be moved, or be withdrawn without notice, and no pricing was published because there was none to publish.

Who it is for

It is aimed at people looking for conversational guidance on workplace and career questions from the perspective of an expert they recognise. The framing is closer to a self-help conversation than to a search engine, and that shapes expectations: you come for reflection rather than a definitive answer.

What to keep in mind

Availability is the concrete issue, since the listed page is gone. Treat the entry as a pointer to an idea rather than a working service until you find where it currently lives.

The more important caution is about what you are talking to. The representation is AI and not the person, so treat the advice as generated rather than as the expert speaking, and never assume confidentiality for anything you type. An experiment at this stage typically has a short privacy notice rather than a mature retention policy, and a career question often contains information about your employer, your manager or your pay that you would not put in a public form.

There is also a consent question on the other side of the conversation, which resolves in the operator's favour rather than the user's: the people being modelled are named individuals whose likeness and voice are reproduced, and whether that arrangement is disclosed and authorised is a matter for them and the platform. A user cannot verify it, which is a reason to treat the output as entertainment with useful prompts rather than as the endorsement of a named person. It is also worth noticing what the format does well, which is to make a familiar line of advice feel conversational enough to act on: a career suggestion you have read a dozen times lands differently when you have to answer it. That is a real benefit of the interface, and it does not depend on believing the voice belongs to anyone.