Your client is going to walk into the meeting with an AI in their ear

Peter Diamandis published a dated prediction this week: in three to five years you will give your AI permission to know everything about your life — schedule, blood tests, conversations, stress level. He calls this ambient AI, and says it arrives around 2028.

The newsletter describes five scenarios, almost all domestic: the robot that prepares breakfast by checking your glucose, the car that's already at the door when the meeting ends, the house that changes the lighting because you arrived tired. These are consumer illustrations, and it's easy to read them as distant futurology.

But there's one scenario on the list that isn't about your home. It's about your meeting. And that one is already starting.

The scenario that changes your business

In the fifth scenario, someone walks into an investment meeting with an AI in their ear. Before going in, they receive a summary: name, recent investments, the sector the person is betting on, the sector they just turned down. During the conversation, the AI detects from the tone that the other party is skeptical about the timeline and whispers: "address the timeline objection now, he's comparing you to a company that missed its schedule last year." Afterward, it has already drafted the follow-up email.

Diamandis describes this as an advantage for the user. Turn the tables: this is how your client is going to arrive to negotiate with you.

They'll sit down already knowing your average price, what your former clients complained about, how much the competitor charged for something similar, and how long you usually take to deliver. Not because they researched all night, but because they asked once while driving.

This isn't 2028. It's now, in a simpler version

The part with the robot and the body sensor really is the future. But the part that affects your sale already happens every day, and most companies haven't noticed yet:

• The client asks ChatGPT "which companies do this in my city" before opening Google.
• They paste your proposal into an assistant and ask it to compare it with another, item by item.
• They ask for a list of questions to ask you — and it's usually better than the one they'd come up with alone.
• They check whether the price you gave is in line with market rates.

Anyone who sells to companies has already felt this: the quality of questions from the other side has gone up, and the client arrives with a repertoire that doesn't match the size of their operation. They didn't get smarter. They got better advised.

What changes in practice for sellers

1. Improvising no longer works. It used to be possible to adjust the price based on how the client looked, because they had no reference point. Now they do. Anyone working with numbers made up on the spot will get caught, and getting caught once costs the whole sale.

2. Hidden information became a disadvantage. The company that doesn't publish what it does, how it charges, and what it doesn't do simply doesn't enter the AI's answer. It's not that it comes up second: it doesn't come up at all. Whoever isn't mentioned isn't compared, and whoever isn't compared isn't hired.

3. The objection arrives ready-made. If there's a public criticism of your type of service, it will be on the table. It's worth having the answer ready — and having published that answer somewhere the AI can read it.

4. The memory of the meeting is no longer just yours. If the other side has an assistant, whatever you promised off the cuff gets recorded precisely. That's good for those who deliver what they promise, and costly for everyone else.

Where Diamandis is talking about a world that isn't yours

He resolves the privacy question in one sentence: "privacy is an old concept, we already leak data all the time, and the trade-off is worth it." As a personal opinion from someone who decides about their own data, that's fine. As guidance for a Brazilian company, it's a costly mistake.

The difference is simple: he's talking about you giving up your own data. Your company handles third-party data — client, patient, employee, supplier. In that case, you're not the one deciding, and LGPD (Brazil's data protection law) doesn't ask whether the trade-off was worth it. Three concrete consequences, once AI starts listening:

• Recording a meeting with a client requires notice. An assistant that transcribes on its own counts as a recording. Giving notice isn't a formality, it's the legal basis.
• A transcript is personal data. Once stored in any tool, it needs a purpose, a disposal deadline, and someone responsible for it.
• It's not okay to upload client data to "optimize." The legal basis you have to serve that client isn't the same one that authorizes feeding what they told you into a third-party system.

That's not a reason to stay out of it. It's a reason to go in with clear boundaries defined — and that's exactly the subject of the permission boundary nobody defines.

What to do in the coming months

None of this requires new technology. It requires that the right information be written down:

• Publish what the AI needs to cite. What you do, for whom, how you charge, what you don't do, and how quickly you respond. A vague page doesn't turn into a citation — and that's exactly what AI optimization solves.
• Write your objections before the client does. The five hard questions you hear every week, answered in public text. They start working for you while the client talks to their AI.
• Standardize price and timeline. You don't need to publish a price table; you need your answer to be the same in January and in July.
• Record what you promise. If the other side has perfect memory, having yours too stops being bureaucracy.
• Define your data boundary before you need it. What can be recorded, what can be transcribed, where it's stored, and for how long.

The conclusion the newsletter doesn't draw

Diamandis closes by listing opportunities for the entrepreneur, the executive, the investor, the student, and the parent. Missing is the list for whoever runs a small company and sells to people — which is most of who will read this in Brazil.

For that person, ambient AI doesn't arrive as a robot in the kitchen. It arrives as a client better informed than you're prepared to receive. And preparation isn't about buying a tool: it's about no longer depending on the client not knowing.

Those who have always worked with clear information will benefit from this shift. Those who relied on asymmetry will feel it. The good news is that leaving the second group behind costs a few afternoons of work writing down what you already know by heart.

Source

Based on the newsletter Auto-Magical Living, by Peter H. Diamandis, published on August 12, 2026 on Metatrends. The five scenarios and the 2028 prediction are his; the reading for small and medium-sized Brazilian businesses and the section on LGPD are ours.