Did We Do This
to Ourselves?

Angie Jones, we meet at last.

Yes, yes. I am honored, long time fan.

Thank you so much and the same back to you. We've been in the same orbit, but on opposite sides of the planet for a long time. Well, welcome to Still Burning. The theme is geeks who still care and are still doing something about it. I like the GeePaw Hill definition of a geek as someone who's highly technical, highly creative, and highly desirous of being both.

I like that.

So here we are in this time of great change. Usually, we record in front of a fire pit, but we're here at a conference about IT leaders in the face of the AI revolution. You gave a presentation yesterday that made me cry. You were talking about the unexpected consequences of AI. Can you talk a little bit more about that?

Yeah. I think that all of us are trying to figure this thing out, and that seems like the very responsible thing to do as leaders. We want to empower our people. We want to teach them how to use this new tool that's non-deterministic to solve engineering challenges. And I made really good progress at doing that within a large engineering org. And once we got to the stage where everyone is running five, 10 agents at a time, and we're automating all of this work and delegating tasks to the agents, the rug was pulled out. And with those efficiencies, then came a large cut. And we see layoffs across the industry quite often these days.

It's been going on for a couple of years.

Yeah. This one was a 40% cut that was cited as an AI rift. That left me with the question of what are we doing? Is this what we're aiming to do?

Are we working ourselves out of a job?

Yeah.

And who benefits?

Who benefits from that?

So, yeah, and I don't think there's an answer for it. It's not like we can stop this, but how do we do it responsibly? How do we reduce the harm? How do we maximize the benefit — optimize the benefits — while reducing the harm.

And I thought the presentation was a beautifully crafted presentation. In that you built to this, and you were the difficulties, and we overcame, and then you dropped the mic. And asked those critical questions — like, did we do this to ourselves?

Right.

Was there an alternative?

Yeah, and I think we have to stop lying to ourselves. We tell folks, oh, no, you have to do this. If you don't, you'll be left behind — which, I mean, that's true. But also, we give some false sense of you'll be safe if you do do this, right? And that has not been true in my case.

Yeah, absolutely. So it's damned if you do and damned if you don't.

That's right.

If you don't adopt these tools, then you'll be fired. And if you do adopt these tools, then you'll be fired. And nobody seems to be doing the math on that and saying, is there a different bargain that we could offer?

Yeah, and I'm curious about that because for years, every business has said, if only we had more people, we could do all of these amazing things. And so now we're in a position where you don't necessarily need to hire more people, but you can maybe 5x, 10x the people that you have. Why aren't we looking at — OK, now all of those things that we said we wanted to do, let's do them. Is that a possible outcome?

Well, is the market going to sort this out? Are there going to be organizations that figure out that there are better uses for people than dropping their salaries? It seems like the assumption is: the best use we have for people is eliminating their salaries. Really — there's nothing else you could do with all of the passion, the experience, the connections, the social relations? That's the best thing you can think of to do is stop spending money on them.

So shifting gears a little bit, how did you first know that you were a geek?

Probably. So I'm not one of those folks who grew up with a computer and learned to hack. I didn't find computer science until I was already in college, very unsure of what I wanted to do with my life. And my father, who was an accountant, said, hey, look, these computers are about to take off. You should learn how to use one. And me not knowing anything about computers, I just find the first class that has computer somewhere in the description. And that was a C++ course.

Oh, I'm so sorry.

But that was the best thing that happened to me. This was fascinating to me. I always loved solving puzzles and things like that. So this felt very much like that. And so I did really well in that class. And that's when I realized — when class lets out and everyone's going on the yard and doing fun college things and I'm running to the computer lab because I can't wait to do my program and assignment on my little floppy disk — that's when I realized, OK, I might be a geek.

And now after this, you've got a new position at a new organization. Can you tell me a little bit about that?

Yes, I'm really excited about this one. So this is at the Agentic AI Foundation, which is a new foundation founded by Block and Anthropic and OpenAI as part of the Linux Foundation. And so this one focuses on key standards and primitives in the agentic AI space. I think this is really important because we're seeing a lot of innovation from companies everywhere and everyone is trying to solve the same problems at the same time. Companies need some assurance of which standards can I build on top of? Which ones will stick around? Which ones will not be locked inside of one vendor? And so the Agentic AI Foundation serves as this neutral governance for key primitives like MCP, for example, AgentMD. I think of it as like the grown up job where all of these labs are kind of fighting and racing. Come on, let's sit at the table. Come play nice for a few minutes and let's figure this stuff out.

Got it. So your job now is only partly technical and it's partly getting these people to work together. Do you carry a set of principles into that conversation?

Yes. I think I've always advocated for developers.

Yeah, I called you the geek whisperer.

I'm the geek whisperer. Right. And so my work prior to this in enabling these developers — we had all the tools, we had everything. And so I know firsthand how difficult it is to build on top of this stuff, how to use it. So I bring that with me. The interoperability of it all — these are things that are at the forefront of my mind when having these types of conversations.

And so there's a balance to be struck between specifying, locking some protocol down too early and diminishing innovation, and locking it down too late — now you've got VHS and Beta. So how do you navigate? Do you have a gut sense of when the timing is?

Yeah. I just announced that we're now open for project submission. And the criteria I specified is that it's mature and has adoption. The community's response: "What do you mean mature? Nothing's mature right now." And so, you know, this is a really delicate balance. All of this stuff is new. It might work today. Who knows if it'll work tomorrow? And so there needs to be some sense of adoption, like some sense of interest that developers, companies, people want this. They need it. They want to build on top of it. I think MCP is our poster child of that. That's a very young protocol, and yet the community demanded — they are the ones that said, this is the one that we're going to go with. And so that became obvious.

Is there something that you find intriguing that's come out recently?

I am. The computer use stuff is very intriguing.

Oh, same. Same. And also, it seems both so unsafe, but also has so much promise.

Right. And this is another one where the community is saying, no, we're going this way. Figure out the safety and the security and everything along the way. Like this is where we're going. And so I feel like we're almost in a reactive mode with all of this stuff. Everything is just kind of exploding. And then the grownups have to run around and kind of put the patches in place and the guardrails to help people not hurt themselves. Agent skills is also one that I'm fascinated by. You'll hear people minimize this — it's just a markdown file. But you can do some really powerful things with this setup.

Yeah. I'm interested in exploring that also. So how are you specifying the standard? We're talking about harnesses, which just sounds like a test suite to me.

We aren't the ones that specify that. Another company or individual would be the ones that come up with that. And then we just house that for them once they figure out the detail. MCP, for example — Anthropic is the one that specifies what's in and what's out. And they have a community that they utilize to help refine that spec. And that's not just Anthropic. That's multiple employees as well. We kind of let folks do their thing.

So back to Claude just for a second. What do you want to build with it?

I want to build everything with it. What's interesting is I went from a mature company where processes and things were already in place, and now I'm at Agentic AI Foundation, which is brand new. So I'm like — before we hire anyone or before I use any of the services that are already existing at Linux Foundation, I want to think about this from an AI-first mindset and build this foundation AI natively. Pretty much everything I want to at least try to see how much of this can I agentify, and then hire the people for the things I need the people for.

I always love that self-description loop. Because if you're using it in-house, you have the shortest possible feedback loop.

So back to the junior developer question. How do we get them to be senior developers? And what does senior developer even mean in this new world? You know, parents come to me and say, I have a junior in college, three years into a CS degree, and now they're worried they'll never get a job. And I just think: you're a carpenter. You've just seen your first circular saw. And now you're saying carpentry's over because you don't have to saw by hand? There are so many more things you'll be able to address by writing programs than you've ever been able to do before.

Yeah. I actually launched this program at Block called AI Builder Fellowship. I brought in 10 engineers. Some had a design focus. Some were hardcore software development. And my idea was — these folks don't have the same biases and muscle memory that we have. They are native to these tools. This is how they have learned to build software. I embedded them into teams to do real product work. And just as you said, the seasoned engineers were able to learn from them. If you see someone come in off the street, first job, and all of a sudden they're putting up more PRs than you are — that's really uncomfortable. Right? And so two things: you either become more curious about how are they able to do this, hopefully learn something from it, or two, you're going to try to break their habit. But it's a two-way street.

Even the way that I designed that interview loop was much different. I said, bring your agent of choice. Here's some money for tokens. Then the interview process was: hey, here's a fairly open-ended challenge — build a learning tool for something you would want to learn. And this gave us so much insight into how they prompt, how they engage with the agent — are they just taking the output blindly or are they taking a moment to look through it and verify it's on the right path? And it's OK if the genie messes up. That's expected. But how do you handle that? Are you completely lost or do you know how to rein that genie back in?

Yeah, I tend to think that curiosity is the natural state and we pound it out of them. Whether it's through organizational structure — don't spend so much on tokens, don't touch that part of the code. You know what — when you had these juniors talk directly to customers, magic happened.

Exactly.

So why don't we want that magic to happen?

We should want that. Yeah.

Closing question. What scares you about this?

Everything scares me about it. Beyond just one job or one company, looking at the industry as a whole, the world as a whole — I've spent some time, some hours laying on the floor, looking at the ceiling. Where are we headed as a society? What happens if the work that humans have been doing is no longer a human job? What happens to society? And so that's a really scary thought. I can't stay there too long. But I circle back on it often. And I don't have any answers. I told my genie one of those times I was laying on the floor, I picked up my phone. I'm like, don't bullshit me. Where's this all going? What are the humans going to do? And this was the one time it did not say, oh, no, we'll always need humans. It was like, they don't know. They don't know.

What a perfect place to end, Angie. What a pleasure getting to know you a little bit. And with that, it's still burning.