Tools & Resources Archive Details

Google Labs Portraits

What it is

An experimental Google Labs product for conversational “AI coaches” based on real experts (starting with Kim Scott) that uses Gemini to answer in the expert’s voice and domain.

Gabriel’s notes

Portraits is an experimental product from Google Labs that lets you chat with AI representations of “trusted experts” built in partnership with the experts themselves. Google’s launch write-up says the Portrait uses Gemini to generate responses that draw from the expert’s content and speak in their voice via an illustrated avatar, and that it’s available in the U.S. for users 18+ (at least at launch, the first featured expert is Radical Candor author Kim Scott). If you’re looking for “mentorship vibes” without scheduling a human, this is the whole point.

Quick take: I talked with the Kim Scott “Portrait,” and it’s one of the better implementations I’ve seen of the “AI coach” idea—useful, focused, and (crucially) constrained to a real person’s body of work. It feels like the wholesome version of “talk to your heroes,” with fewer chances of your hero turning into a chaos gremlin.

I saved this under AI because it’s a clean example of a practical, narrow-ish use of generative models: turning an expert’s existing corpus into an interactive coaching interface (instead of yet another everything-bot that confidently tells you to eat rocks).

My original note, cleaned up: I had a conversation with the AI version of one of my non-fiction author heroes—Kim Scott—using Google Labs’ experimental tool, Portraits. It was genuinely impressive, and I’d recommend trying it if you want coaching-style guidance grounded in a specific expert’s frameworks.

Good fit if you want to:

  • Practice tough-work-conversation scenarios (feedback, conflict, “help, I have to talk to a human”).
  • Get leadership/management coaching that stays close to one expert’s ideas instead of drifting into generic advice soup.
  • Explore how “expert-backed” AI experiences differ from community-made personas.
  • Prototype internal training prompts and see what employees might ask a coach—before you build your own.

Pricing snapshot (auto-enriched):

Unknown / not confirmed. Google’s announcement materials I reviewed don’t clearly state a price. Some third-party directories report it as free during the early experiment phase, but treat that as provisional until Google states it explicitly.

Work-use / compliance snapshot (auto-enriched):

Portraits is a Google Labs experiment (U.S., 18+ at launch). Your use is generally subject to Google’s Terms and Privacy Policy. Unknown / not confirmed whether Portraits conversations are used to train or improve models in a product-specific way, so assume anything you type could be logged/retained per Google’s standard practices and your account settings. For work use: don’t paste sensitive company info, client data, passwords, or anything you wouldn’t want showing up in a discovery request.

Alternatives (auto-enriched):

  • Character.AI: Great for chatting with persona-style characters, but Portraits is positioned as “built with the expert,” not just a convincing imitation.
  • ChatGPT (with a well-designed system prompt / custom GPT): More flexible and general-purpose, but it won’t automatically have a sanctioned, expert-provided voice/corpus unless you build that structure yourself.

Before you adopt it:

  • Run a “prompt audit” on yourself: what questions are you tempted to ask that you shouldn’t (legal/HR, confidential, medical)?
  • Test with a fake scenario first, then rewrite your real scenario in sanitized form.
  • Capture outputs as draft language and apply human judgment—especially if the advice affects people’s careers.

Sources

  • https://blog.google/innovation-and-ai/models-and-research/google-labs/portraits/
  • https://www.androidauthority.com/google-portraits-gemini-3564749/
  • https://www.techradar.com/computing/artificial-intelligence/i-tried-portraits-googles-new-way-of-turning-real-life-experts-into-your-own-personal-ai-life-coach
  • https://policies.google.com/privacy?hl=en-US

Visit the resource