Free live masterclass for product managers

How AI Actually Works, in Plain English

In one live session you’ll learn how a language model works, why the same prompt gives you different answers, why it makes things up and what each request costs. By the end you’ll be able to explain it on a whiteboard in your next interview.

Friday, October 9
12:00 PM ET / 9:00 AM PT
Live online, about 75 minutes

It’s free, and you won’t need a credit card.

By the end of the hour, you’ll be able to explain:

1
How a model turns your prompt into an answer, one token at a time. We draw it on a whiteboard, with no math.
2
Why the same prompt gives a different answer each time, and how much of that variation your product can live with.
3
Where made-up answers come from, and the places in your product where a confident wrong answer would hurt.
4
What one request costs. We’ll price a real call on screen, then work out the bill at 1,000 users and at 100,000.
5
How to pick the right technique for a problem, and how to tell when AI is the wrong tool.

Then we build one, live

We’ll build a meeting note‑taker from scratch in Claude Code while you watch.You’ll see a real transcript go in and the decisions, action items and open questions come out.Then we put it through three tests.

1
Run it twice
Same transcript, same prompt. You’ll see how much the answers change.
2
Price the call
We’ll work out what each run costs.
3
Break it on purpose
We’ll make it give a confident wrong answer, then fix it the way a PM would.

Walk into your next interview ready for the AI questions

We’ll finish with questions candidates say they were asked in PM interviews at OpenAI and Microsoft, including these two. You’ll get an answer outline for each one, built from what you saw in the session.

Questions compiled from candidate reports on Exponent, IGotAnOffer, Glassdoor and Blind, July 2026.

“Define hallucinations in LLMs.”

“Explain the trade-offs in model selection.”

This session is for you if:

You use ChatGPT or Claude every day but couldn’t explain how it works.
You’re a PM adding AI to your product, and you want to make better calls with your engineers.
You’re moving into product management and AI keeps coming up in your interviews.
You’re an engineer moving into product and want to understand the PM side of AI.

You’ll leave able to talk about AI fluently. You won’t write any code, and an hour won’t turn you into an ML engineer.

Why this matters now

12,397

AI product jobs were posted in the US in the first half of 2026. Companies want PMs who can explain what’s going on under the hood.

Source: Axial Search, H1 2026.

A bonus gift for everyone who stays until the end of the webinar

The exact prompt we use to build the note‑taker
A sample meeting transcript to practise on
A one‑page cheat sheet on tokens, probabilities, context and cost

Your presenter

Aakash Gupta

Former VP of Product at Apollo.io

1M+ followers across social platforms

Aakash runs Product Growth, a newsletter and podcast for product managers. Before Apollo.io he led growth at Affirm. He builds his own AI tools in Claude Code every week.

Previously at