What no one tells you about building AI features (until it's too late)
AI features are exploding in popularity, but building one that actually works for users is harder than it looks. The technology is new, the expectations are sky-high, and the rules are still being written.
The difference between a flashy demo and a product people trust comes down to a few critical choices: how you design for transparency, how you set (and reset) user expectations, and how you handle the weird, unexpected moments when your AI goes off the tracks.
In this talk, we’ll unpack key lessons, learned from working on two worldwide, flagship products: Microsoft Excel and Copilot - that will turn your AI from “just another feature” into an indispensable tool. You’ll hear what works, what doesn’t, and how to skip the pitfalls that sink most products before they scale.
No matter where you are in your AI journey, you’ll walk away knowing how to make a product people actually want to talk to - and talk about.
AI features are exploding in popularity, but building one that actually works for users is harder than it looks. The technology is new, the expectations are sky-high, and the rules are still being written.
The difference between a flashy demo and a product people trust comes down to a few critical choices: how you design for transparency, how you set (and reset) user expectations, and how you handle the weird, unexpected moments when your AI goes off the tracks.
In this talk, we’ll unpack key lessons, learned from working on two worldwide, flagship products: Microsoft Excel and Copilot - that will turn your AI from “just another feature” into an indispensable tool. You’ll hear what works, what doesn’t, and how to skip the pitfalls that sink most products before they scale.
No matter where you are in your AI journey, you’ll walk away knowing how to make a product people actually want to talk to - and talk about.
Got some juicy gossip?
We’d love to hear it (and any other questions, wishes or suggestions you have).
