The AI Spend Question Nobody Can Answer

**SPEAKER_1** (0:00)
This episode is brought to you by Accenture. When you're advertising operations fall out of sync, everything else follows. Spotify and Accenture are working together to reinvent the rhythm of ad sales. Using automation, analytics, and smarter workflows to simplify campaign delivery and access better data across the business. The result? Less time spent on operations, more time connecting brands with the moments and fandoms that matter most. Learn more at accenture.com/spotify.

**SPEAKER_2** (0:31)
When you need to build up your team to handle the growing chaos at work, use Indeed Sponsored Jobs. It gives your job post the boost it needs to be seen and helps reach people with the right skills, certifications, and more. Spend less time searching and more time actually interviewing candidates who check all your boxes. Listeners of this show will get a $75 sponsored job credit at indeed.com/podcast. That's indeed.com/podcast. Terms and conditions apply. Need a hiring hero? This is a job for Indeed Sponsored Jobs.

**Amos Haviv** (1:00)
We're in the 4,000 engineers. We're operating 8,000 repos. We have new code contributors that did not use to contribute before, because everyone is an engineer these days. Everybody's burning through their budget. Uber and Microsoft. How do you control the budget? How is the budget being managed? The Internet was a bubble. Did the Internet go away?

**Patrick Akil** (1:18)
I love the Internet.

**Amos Haviv** (1:19)
Now when people ask us, do people that use AI deliver more features? We can say yes or no.
He's AI turning us all into managers.

**Patrick Akil** (1:27)
So when I met you and when I went to your talk, I was thinking, this guy's pretty far. In the whole agentic engineering journey, you've solved a lot of problems, and I want to zoom into some of the problems you solved and how.
But right now to start, what is the biggest problem you're trying to solve right now?

**Amos Haviv** (1:47)
I think like everyone, it's a cliche at this point.
Verification, specifically code reviews. It's something that a lot of the people in the industry are struggling with. But as a whole, I think that generalizing from that problem is the next frontier, meaning the bottlenecks are going to move in other areas of your STLC. Yeah. So right now it's a with code review. I don't know, next week it should be like with security.
You have released pipelines that are also still deterministic in old school. So you will have to figure release again. So I think what we're trying to do is generalize, and come up with a way to say, okay, so we solve for code review. Let's say we do.
What is the next bottleneck? Can we predict it? Can we be there before our engineers are hitting the wall? And I think that's what we're trying to do right now. Because it's obviously coming.
I think if you ask me, I think some of the bottlenecks can go up upstream. Like, can we generate enough business?
I think there's a very good book that I really like. It's been 43 years, I think, since it was written. It's called High Output Management by Andy Grove. And the metaphor there is, it's funny, he was the CEO of Intel. He got Intel to where Intel was in like 80s and 90s. Wow. And he talks, his main metaphor is a diner. Like how to make breakfast. And he says, it doesn't matter how quickly you toast the bread. It doesn't matter how quickly you cut the salad. If your eggs take 10 minutes to boil, then your breakfast will take 10 minutes to get out. And that is something we're seeing right now. So it doesn't matter how quickly our engineers can code, or how quickly can we do code reviews. If we can, for instance, produce enough designs or PLDs or high quality business heuristics, then we won't get anything more quickly out there. Gotcha.

**Patrick Akil** (4:13)
So it's really a shared operation. It all needs to work quicker to get the whole thing out.

**Amos Haviv** (4:18)
Yeah.
It must be like that. We are as slow as our slowest link.

**Patrick Akil** (4:23)
Gotcha.

**Amos Haviv** (4:25)
And so we are all depending on each other, which is kind of nice as well. It's like there can be no blame culture anymore, because we're all in this together and we're all trying to operate, and button X will move around in companies. So yes, right now it's code review.

**Patrick Akil** (4:41)
Before we go into how you're solving that, how you're tackling this code review problem, can you share a little bit more about the scale of the operation? How many engineers?