Fireside: hiring for AI product teams · Notes

The short version

Farah walked through how Globex rewrote its PM hiring loop for AI work: no more whiteboard case studies, just a live review of real model outputs. Her view is that curiosity and clear writing now predict more than years of experience.

4 takeaways

Key takeaways

01Globex replaced the take-home case study with a 45-minute live session where candidates review real agent outputs and decide what ships.
02Years of experience predicted less than they expected. Curiosity and clear writing predicted more.
03Every new PM pairs with an engineer on an eval set in their first month.
04Don't hire for an "AI PM" title. Hire good PMs and give them real time with the models.

6 chapters

Chapters

Timed to the recording. Tap a time to link to it.Read each chapter in full

0:00
Why the old loop broke
Case studies rewarded polish. The job now needs judgement on messy, probabilistic output.
5:20
The live output review
Twenty real outputs, forty-five minutes, one question: which of these would you ship, and why.
12:40
What predicts success
Curiosity, clear writing and a habit of building small things. Tenure mattered less than they thought.
19:30
The "AI PM" title
Why Globex dropped the title, and how they level PMs who work on model features.
26:00
The first 30 days
Pairing on an eval set, shadowing support, and shipping one small fix before any roadmap work.
31:10
Questions from the room
Hiring juniors when agents do junior work, and how to interview designers for AI features.

3 quotes

Worth quoting

We stopped asking people to design a product on a whiteboard. We hand them twenty real outputs and ask which ones they'd ship.
Farah SiddiquiVP Talent, Globex Payments6:05 in the transcript
Ten years of experience can be one year, ten times. I'd rather see what you built last month.
Farah SiddiquiVP Talent, Globex Payments14:30
There's no such thing as an AI PM. There are PMs, and some of them have spent real time with the models.
Farah SiddiquiVP Talent, Globex Payments20:45

Transcript excerpt

From the transcript

An excerpt, lightly cleaned up. Speakers were matched by voice, then checked by a person.

5:20Kabir Shah
So walk me through it. Someone comes in for a PM role at Globex. What happens?
5:31Farah Siddiqui
No whiteboard. We sit them down with a laptop and a real feature, anonymised, and the last twenty things the model produced for customers.
6:05Farah Siddiqui
We stopped asking people to design a product on a whiteboard. We hand them twenty real outputs and ask which ones they'd ship.
6:40Kabir Shah
And what are you listening for?
6:46Farah Siddiqui
Whether they slow down. The good ones get suspicious of the outputs that look fine. The weaker ones skim.
31:15Audience
If agents do the junior work, how do juniors ever learn?
31:24Farah Siddiqui
Honestly, it's the question I lose sleep over. Our answer for now is that juniors review the agent's work before seniors do. They learn by catching its mistakes.

The short version

4 takeaways

Key takeaways

01Globex replaced the take-home case study with a 45-minute live session where candidates review real agent outputs and decide what ships.
02Years of experience predicted less than they expected. Curiosity and clear writing predicted more.
03Every new PM pairs with an engineer on an eval set in their first month.
04Don't hire for an "AI PM" title. Hire good PMs and give them real time with the models.

6 chapters

Chapters

Timed to the recording. Tap a time to link to it.Read each chapter in full

0:00
Why the old loop broke
Case studies rewarded polish. The job now needs judgement on messy, probabilistic output.
5:20
The live output review
Twenty real outputs, forty-five minutes, one question: which of these would you ship, and why.
12:40
What predicts success
Curiosity, clear writing and a habit of building small things. Tenure mattered less than they thought.
19:30
The "AI PM" title
Why Globex dropped the title, and how they level PMs who work on model features.
26:00
The first 30 days
Pairing on an eval set, shadowing support, and shipping one small fix before any roadmap work.
31:10
Questions from the room
Hiring juniors when agents do junior work, and how to interview designers for AI features.

3 quotes

Worth quoting

We stopped asking people to design a product on a whiteboard. We hand them twenty real outputs and ask which ones they'd ship.
Farah SiddiquiVP Talent, Globex Payments6:05 in the transcript
Ten years of experience can be one year, ten times. I'd rather see what you built last month.
Farah SiddiquiVP Talent, Globex Payments14:30
There's no such thing as an AI PM. There are PMs, and some of them have spent real time with the models.
Farah SiddiquiVP Talent, Globex Payments20:45

Transcript excerpt

From the transcript

An excerpt, lightly cleaned up. Speakers were matched by voice, then checked by a person.

5:20Kabir Shah
So walk me through it. Someone comes in for a PM role at Globex. What happens?
5:31Farah Siddiqui
No whiteboard. We sit them down with a laptop and a real feature, anonymised, and the last twenty things the model produced for customers.
6:05Farah Siddiqui
We stopped asking people to design a product on a whiteboard. We hand them twenty real outputs and ask which ones they'd ship.
6:40Kabir Shah
And what are you listening for?
6:46Farah Siddiqui
Whether they slow down. The good ones get suspicious of the outputs that look fine. The weaker ones skim.
31:15Audience
If agents do the junior work, how do juniors ever learn?
31:24Farah Siddiqui
Honestly, it's the question I lose sleep over. Our answer for now is that juniors review the agent's work before seniors do. They learn by catching its mistakes.

Key takeaways

Chapters

Why the old loop broke

The live output review

What predicts success

The "AI PM" title

The first 30 days

Questions from the room

Worth quoting

From the transcript

Key takeaways

Chapters

Why the old loop broke

The live output review

What predicts success

The "AI PM" title

The first 30 days

Questions from the room

Worth quoting

From the transcript