Building the Foundation for the Agentic AI Era

0:02Welcome to the Practical AI Podcast, where we break down the real-world applications of artificial intelligence and how it's shaping the way we live, work, and create. Our goal is to help make AI technology practical, productive, and accessible to everyone. Whether you're a developer, business leader, or just curious about the tech behind the buzz, you're in the right place. Be sure to connect with us on LinkedIn, X, or Blue Sky to stay up to date with episode drops, behind-the-scenes content, and AI insights. You can learn more at practicalai.fm. Now, on to the show.

0:41Hey, welcome to another edition of the Practical AI Podcast. I am your co-host going solo today. I'm Chris Benson. Daniel's not with me this time. but we have an excellent conversation coming up for you. With me today, I have Angie Jones, who is the vice president of the Agentic AI Foundation, which I think is a super cool title to have at a super cool name place. And I'm really looking forward to finding out more about it. Angie, welcome to the show. Thanks so much, Chris. So like I said there in the intro, and I said it before the show started and stuff, if somebody in AI was looking for a play, like the agentic, I mean, that is the coolest sounding thing you can have.

1:29But before we dive too far into the foundation, I'd really like to kind of hear, like, how does someone, like, how do you develop in your career so that you end up doing that? Like, could you tell us a little bit about your background because that's one of those things, if you went and just said something to, like I work with tons of people working on genetic AI, but like leading the foundation would be cool. So I'm just curious, like how do you get to that point? Yeah. So I'm your traditional techie. So I've worked as an engineer for a couple of decades. So, you know, at places like IBM and Twitter.

2:11and the last role was at Block in an engineering leadership capacity. And so in that role, one of my tasks was to basically teach the entire company, 12 ,000 people, how to use AI agents. And this is as I'm learning myself, because I mean, there's no book for this, you know, I'll write out the gate. And that was early like 2024. So a lot of this stuff was brand new. It was not even common in tech, let alone in others' verticals, right? And so I also lead developer relations. And so a big part of my role is helping developers worldwide understand new technologies and how to use them. And so our company, Block, created this internal AI agent named Goose.

3:03and Goose, the developers were using Goose to like automate engineering tasks, help with coding and things like that. And so we were teaching developers across the globe, what is an AI agent? Like we were very early in this. And so Jack Dorsey, who leads Block, was like, hey, Angie, you're teaching like everybody else about agents. I really would love for everyone in this company to learn how to use agents. And so I'm talking finance, marketing, design, HR, everyone needed to learn about AI and more specifically how to utilize agents. And so me and the team went really deep on this. And then it got to a point, pretty AI fluent company where everyone is comfortable using this.

3:56I needed to go really deep on the engineering org. And so that was my home. Um, our engineers were using this, but we weren't seeing a big difference in like developer velocity, for example. Right. And so we, we explored that and learned that, you know, we're only at the tip of the iceberg. We really could do a whole lot more to get to this autonomous engineering org. Right. And so I pretty much drank from the fire hose of like all of the news, all of the releases, everything that's going out. and I consume that, kind of filter out a lot of the noise and then bring the things that are valuable to the engineers.

4:39And so I would say like, I know a lot about the space. And also at my time at Block, I worked on a protocol. So this was like a cross border money movement protocol. So those who don't know Block, that's the company, the finance tech company that houses Square and Cash App, right? So money is our jam. So I worked on this protocol, never thought I would be a protocol girl, but learned a lot about like just kind of open standards and how all of that works as well. Also really big and open source throughout my career. I've always like believed in open source, contribute to open source, like advocate for it.

5:19Right. So all of that kind of came together in this perfect storm as OpenAI, Anthropic and Block wanted to form a foundation for agentic AI. They understood that, hey, some of these standards, some of these open source projects that we're coming up with, we probably shouldn't be the sole authors or owners of these things. Right. MCP is a great example. So MCP is the model context protocol. This is what agents use to connect to applications and tools. Right. Which we saw across block. Everyone in block needed MCP service to connect to whatever applications they were using. And so like Anthropic realized, yeah, this probably should live in a neutral home.

6:09And so those three companies came together to form the Agentic Foundation under the Linux Foundation. So Linux Foundation has been around for decades. Everyone knows them. And so, yeah, so this is a new foundation. So once we stood this up, me and our head of open source came over to the foundation. And we thought it was so cool. We started working here full time. Very cool. And I like the Jack Dorsey name drop there. That was that's pretty good. Like, so you actually you were actually working directly with him as well along the way. Yeah, that's right. So I worked with Jack at at Twitter. And then he brought me over to Block when they when they started developing like these open source projects and protocol.

6:58Very cool. I want to back up for a moment because I got a couple of questions from things that that you brought up there. And one is the education, because I think educating the organizations that folks are in, like you were, you know, pathfinding along the same kind of task that a lot of organizations are trying to do right now. And that is, you know, especially like every year recently, but especially 2026 has been just insanely fast in terms of the level of progress and the onset of Agentex. You know, they were there last year, but this year it just has taken over the world. And so like every org is dealing with that now.

7:40And you have taken point on this notion of education, not just for developers, but for, you know, the whole org and kind of parts of the org that maybe people aren't thinking about as much because they tend to be very focused on developers. Can you talk a little bit about what creating that kind of change looks like in terms of, and I'm going to separate it. I want to ask about the non-developers first. Like when you're going into finance and you're going into these other organizational departments that are not thinking about the bits and bytes of AI all the time and you're trying to say, here's a new tool.

8:20And like, how do you approach that? Not only from the upskilling that's required, but also from the kind of the, like getting people to accept it. Because, I mean, you know, you see like people out there, there's a lot of resistance to AI in the general population out there. And so, like, how do you navigate that when you're trying to move the org forward like that? Yeah, these were two very big and different challenges, right? So one is the whole change management of it all. Like you're essentially asking people to think differently and do their jobs differently. these are experts in their domains, right?

9:02Who maybe been doing this a couple of decades and you're like, oh, kind of throw away like your processes and everything. And we want you to do this. Not only that, to your point, resistance, right? Fear, lots of different emotions involved in that. And then the whole technical part of it where like these tools, like I said, this was very early on, 2024, 2025, you didn't have these nice desktop applications. You had a CLI, right? You were copying JSON to like get an MCP server working. Like this was foreign to folks who like don't use these tools every day. So it was a really big job. And so like you have to consider all of that.

9:50It's not just, hey, let me show you how to use a terminal, right? It's helping them understand where they still fit into the process, right? With things, it's okay to let that go. Like you probably should not be doing this anymore because it's not a good use of your time anymore, right? And then also, I like to say there's people on both spectrums. There's people that, you know, they think AI is the best thing ever and can do all the things. And there's other people who are like, oh, AI is stupid. It can't do anything. I like to kind of be in the middle there where, you know, I can be realistic about its strengths, its weaknesses, what I should use it for and what I shouldn't use it for.

10:34And just leading with that helped a lot because I'm not trying to get you to drink Kool-Aid, right? I'm saying, hey, here's a different approach that can speed you up, right? That can also give you access to a lot of stuff that maybe you didn't have before. And that part was the key. So when there's, let's say data, right? Everybody needs data, no matter what role you're in. If I say, hey, listen, I can get you access with this agent and like some MCP server to, let's say, records that are in this database that you would have had to go to another team, put in a request and ask for a report, right?

11:17And then try to figure that report out yourself, I can get you that data in seconds so that you can go and like do your best work. So people are like sold on that kind of thing, right? So you take them step by step and give them things that are meaningful to them that might not be their core job. So just so they could get used to it and they could start building some trust, right? And then eventually they start delegating a little bit more and a little bit more until they find that right spot of, okay, I shouldn't delegate this. The AI does not do well with that. That part I'll take, right? So that was my overall approach to it.

11:57I really like that. And I know when I get in conversations with folks about this topic, one of the things that I often say, which I think is kind of resonating with your story there, is that find the place where the human has a distinct advantage from the AI and differentiate the AI. You know, go down through the job requirements and find where the human is still better at this point. And that may be fluid. That may change over time depending on that. So having an open mind is good. But I think that's incredible advice that you're giving people in terms of how to navigate. Because I think that's one of the biggest questions that people have out there based on the conversations I'm having.

12:42To flip to the other side of the coin when you're dealing with developers. And I'm a lifelong developer. I've been doing it for decades. and I see, you know, every day on social media and stuff, I see people, you see the developers that are embracing Agenix fully and they're orchestrating. And then you see the ones that are like, oh, it still sucks. And, you know, I'm still, you know, all that stuff. And what was your experience as you dived into the pool of developers trying to get the embrace going here? So for a while, if we look back, So back then, you're like, oh, this is great. But it wasn't that great, right?

13:24I mean, in the early days, it was good for that time. But compared to now, it's not. And so developers who would maybe try the tools and they didn't do a great job, you ask it to write a feature or something, it's like, what? This is stupid, you know? um those developers i found once they tried it once if it didn't work out well they kind of threw their hands up um and it wasn't until like end of maybe 2025 when the like um was it four like sonnet four five or four six i don't even remember the versions anymore but it was like this point in about November of 2025, the models just are really good.

14:13But we couldn't wait until then. Like I had a mandate to like, can everybody use AI and to increase developer velocity? And so what I did, thinking back to that change management, and I, it was too much of a lift to get, it's 3 ,500 developers, to lift everyone at the same time, especially when you have all of these emotions involved and you have this resistance, right? And I would say like we had quite a few people who were resistant to this. So what I did instead, there's this rule. It's called the the one nine ninety rule and this rule essentially says that in any community there's going to be one percent of that community that are creators think about like social media anything like that you'll have nine percent that you know they'll dabble here and there the tinkerers if you will and then the the 90 the bulk of people are consumers right um and so i said if i look at our engineering organization through that lens.

15:20Let me go and put together the one person. And so I went across the org and I pulled people and they didn't necessarily have to be the drunk off the Kool-Aid, like AI peeled people. But I just needed representation from every major repo that we had. You know, our largest ones, our critical ones. I wanted people from various teams, different types of repos as well. So I needed front-end and back-end and mobile, iOS and Android. You know, I needed these people. And so I did this little week-long campaign just going across the org. These people don't necessarily, I'm not their first line manager, right?

16:04So I need to get buy-in from their first line manager. That's a totally different conversation because a lot of them weren't even bought into this, right? But I say, hey, listen, you know, you have this mandate. You got to get people to use AI on your team. Give me one person that I can have 30 % of their time. And what I want to do is one, they all come together. We learn all of this stuff, but it's not just for them. What I need them to do is build that knowledge back into the systems themselves so that your entire team benefits from this, right? And so they were there with me. We drank it, it was 50 of them.

16:46Drinking from the fire hose, trying things out because things are changing so quickly. There's so much noise. You don't know what'll work, what won't work, right? Everything looks great on Twitter in a little demo. But we're working with huge, you know, So monorepos with, you know, hundreds of services in them. Like, you know, we're working with people's money. You know, these are enterprise systems that you have to approach this a bit differently and much more carefully. And so they would try things out. We'll see things. We'll try them. Some things will work for maybe a specific type of repo.

17:27Like, okay, this works great on web. Sucks for mobile, right? We can't use that technique. But together, the 50 of us were able to try a lot of things, figure out what works. Maybe I do have something that works on my iOS repo. Hey, other iOS repos, here's what I figured out, right? And then, like I said, embed this stuff into the system. So things like context engineering techniques, things like, you know, agents MD files and agent skills and all of that baked into the repo so that no matter who was pointing their agent at this repo, the agent could work the way that your team wanted it to work because it was baked into the system.

18:08So we didn't have to, you know, teach everybody how to do this. These champions essentially did that for them and filtered out the noise for their team. So they would bring back the things that actually work that are tried and true. And then they would do a brown bag session or something with their team. Like, let me show you guys what I put in the repo and why it's working for you all of a sudden, you know? So that was the technique that we used there and it worked really, really well. Sometimes as a business owner, it's so hard to put out quality material and content that looks handcrafted and human, but yet still use appropriate tools and automation and AI to make that fast to ship.

18:52That's why I appreciate so much what our partner Framer is doing. Framer is the pro website builder for creators, teams, and businesses that want a professional site and care enough to get every detail right. But that can't come at the price of endless to-do lists and long, long development processes. You need to make that happen fast, which is why Framer has integrated agents directly into their platform that work right alongside the teams creating the site. You can learn more about Framer and learn how to get more out of your site from a Framer specialist or get started building for free today at Framer.com slash practical AI for 30 % off a Framer Pro annual plan.

19:39That's Framer.com slash practical AI for 30 % off. Framer.com slash practical AI. Rules and restrictions may apply. So that is super cool. I'm going to borrow your techniques myself. Okay. I bet a few other folks listening or watching will do the same. Really insightful there in terms of how to approach that. As you arrived, having gone through a lot of this, and you arrive at the Agentec AI Foundation now, and it's brand new as an organization, you're standing it up, you're taking on the responsibilities. Could you talk a bit about like what was it like to get the foundation going and get it focused and get the right people and know that you were doing the things that needed and what those are?

20:32Yeah. So this Genesic AI Foundation was formed end of 2025. So it's been about eight or nine months at this point. Lots of excitement. Right. We had dozens of companies that were interested in being a part because everyone sees the need for this. Everyone is also innovating at lightning speed. There's so much innovation that's happening right now. And if you have anything that you're creating that's like, you know, oh, we don't have this figured out. We need a new standard for this. No one wants that standard being cooked in one kitchen, right? If we're talking about agentic commerce, for example, PayPal doesn't want Stripe figuring that out by themselves.

21:21They want to work together or they need to work together and they know they need to work together. And so what the Agentic AI Foundation has done is provided this neutral home where all of these various companies can come and have these conversations together. So we have working groups, essentially, where the members, they've, OK, which working group do you want to be a part of? And there's so many of them. There's like, like I said, the agentic commerce, right? So if you're figuring out how agents are going to buy stuff on the web, all right, I need, oh, I need Visa, I need PayPal, I need Stripe, I need, you know, all of these various companies together.

21:59And they work on these standards together. I like to think of it as they are defining the rules of the game and creating the game board together. And once they have that figured out, it's like, all right, now we can compete. Deal me in, right? And so that's what the Agentic Foundation does. It also houses those standards or those projects. So right now the projects are MCP, AgentsMD, Goose, AgentGateway, and then also just this week, A2A, which is Google's agent-to-agent protocol, has come into the foundation. And so these projects, now you have companies from all over contributing to these projects, right, or these standards.

22:49You then give it like, you know, more of a voice. If you wanted to use one of these projects, but it lives in one of these companies, you might be a bit hesitant, right? As a company to build your products on top of this, if it lives with this one company, I don't know what they're going to do. I don't know if they'll like kill it. I don't know if they'll just, the roadmap will only reflect their goals. You know what I mean? And so being a part of the Agentic AI Foundation gives it that confidence that, hey, this is in a neutral place. I now can build on top of this. And so that's what it looks like.

23:29We do a lot of education around the projects, the protocols. We throw conferences in all parts of the globe. And I'll say like one really cool aspect of this role. So in my last row, went deep on agents, but I'm also inside of my company, right? A little bit of open source, I'm consuming, but I'm still, my bubble is probably North America. Now this is a global foundation where my job is to pay attention to what is happening across the world. How are they adopting a gigantic AI in Japan, in China, in Africa? You know what I mean? And so now I have this global perspective, which is absolutely fascinating.

24:17And part of the role is also to help these various countries come together on these standards and protocols and projects so that they're interoperable across the board. I'm curious. So that does sound super cool. I'm curious with that global perspective that you have developed in this role, like how do you assess? I think it's very easy for pretty much, I think it's easy for anyone kind of coming from their own perspective and their own little bubble that they're in to kind of assume I'm doing a Gentix and everybody else is doing it just like me or has the same needs. and stuff like that. And so I think it's very easy to forget that diversity of location, diversity of life, all those things can change the user's need for that.

25:12And so do you have any particular highlights from that global perspective on things maybe that surprised you or that caught your interest? I'd love to hear some of that. So I would say in places like China that are mobile first, right? The way they use technology is a bit different than us. So we're in North America, I would say a very like SaaS heavy culture, whereas everything is like mobile over there. And so now like they, for example, they have a big need for a protocol like agent to agent, where you need like this app to be able to talk to that app. And they're both agentic apps, right? And so they could like use that, like for example, like WeChat uses that, you know, with various agents and stuff like that.

26:07So that was really fascinating to me to just see like, oh, even the types of applications you're using essentially influence like the types of tech that are standards or like anything like that, that you might need. And so that's just one example. Yeah. That's cool. I'm curious. So, and this is where I'm going to be selfish on my part. I'm, as an engineer and a research scientist, I'm very focused on autonomy. And as you're getting into mobile and that's getting awfully close to, you know, thinking about edge concerns and stuff like that. And embodied intelligence, you know, is such a hot area.

26:48It's certainly the area that I'm focused on. And, you know, the rapid rise of robotics in all domains, you know, whether they're ground robots or flying things or whatever, is truly taking off like it never has before. No pun intended there. I'm curious how with this kind of, and I think in some parts of the world, you see that more than others. I think in China, Japan, and there are certain places where you're going to see a lot more robotics than you will in the U.S. For U.S. listeners and watchers, we don't have nearly as much of that here. That's right. And so like, how does, how, when you're looking from, from trying to, to take these projects that the foundation has, and you're looking at those different needs and you're saying, well, you know, the Americans and the Canadians and such, you know, kind of have one way of doing it.

27:42And the Chinese and the Japanese and other, you know, like that may have a different way. How do you reconcile, because that's quite different, you know, in terms of how you're using agents in those ways and the way you're configuring it and the way you're constructing them and what their utility is. How do you approach that when you're trying to make everybody happy with standards that are truly meant to be global? Yeah, that is why you need everyone at the table, right? And so if you only had like your top like fame companies in the U.S. kind of determining all of this and there's no one from these other countries that, hey, robotics is like a big deal here.

28:24Like we have to think about the physical applications as well. Right. Then you miss that perspective. Right. And not to say like they don't have places there, but you need leadership from those companies to also have a part in crafting the story. Right. And so that's exactly what the working groups do. Find your lane, get into it. Europe is another good example. So they just rolled out the EU AI Act. And so that is going to change the transparency that these systems have. And that's global. If you're going to be serving anything to someone in Europe, then you have to be able to adhere to these new legislations.

29:10And so that's another example where, OK, I don't care what country you come from. You have to, one, be aware of this and to also help define how this should be done. Right. There was a lot of backlash on how Anthropic wrote this out, for example. Right. And so we have a working group that is looking at, are there some systems that we could build or some standards or something so that everyone doesn't have to reinvent this and figure out how they're going to do it? Right. So then you have the voices from various countries together to figure this out and say, OK, what if we did it this way? They haven't solved that yet, by the way.

29:53But that's an active group that's meeting regularly. And by the way, all of these working groups are open. So anyone can join them. Just, you know, they have meetings, they're public. You could just like kind of go into it, see what they're talking about. Even, you know, chime in with thoughts of your own. But these folks meet and this is, you have one for security. You have one for identity. Like, you know, any lane that you care about, there's someone there like kind of working on this from across the globe. Yeah, I know as a very specific to that, it was just a few days ago, then Theropic released their paper, along with a blog post on watermarking, so that you can detect, you know, AI generated content and all that, which was a fact like if we're, this is probably not the only episode that's going to come up on.