MCPing in the Open with Angie Jones

Justin: [00:00:00] Welcome to Fork Around and find out. I'm your host Justin Garrison. And with me as always is Autumn Nash. And today on the show we are importing some class with Java Champion Angie Jones.

Autumn: I actually like that one.

Justin: Yes. One of the jokes landed. Sweet.

Autumn: I'm so proud of you. Also, like y'all, Angie is the goat. Okay? Like this is like meeting your like superhero in real life. Like Angie is the reason why like I wake up and still put up a tech every day and hope to be her when I grow up. Like.

Justin: Angie, you're currently the VP of DeVere at Block and you've, you're an author, you're a global speaker, you do a lot of stuff in the industry.

A

Autumn: J Java champion Amazing, amazing glasses game like all the time.

Angie: Yes. Thank you for having me.

Justin: Tell us a little about what you are focusing on at Block. What [00:01:00] is it that, I mean, block is is generally known as as kind of finance company, but like you're doing very specific things there.

Angie: Um, I'm doing a lot of things there. Y'all don't see everything I'm doing, but, uh, yeah, I'm a VP of Derel across all of blocks, so that includes our products like Square and Cash App.

Um, it also includes our open source office. So we have, uh, quite a few open source offerings that are pretty popular and we've just released, uh, an AI agent actually called Goose. And so I'm actually just like naturally very excited about Goose. And so that's probably a lot of what you see. And because it's new, I'm, uh.

Kind of, you know, jumping in the ring and playing with it and learning about it. And just like everybody else, I need to learn ai, uh, and how to utilize it in my, in my work, just like, uh, the rest of the [00:02:00] industry. So yeah,

Justin: immediately when I hear Goose, I think of the Untitled Goose game. If you ever played that.

It's like a puzzle sort of game, but a lot of. People that like to do maybe penetration testing. It's kinda like a think outside the box. Like, how are you gonna solve this problem of like, oh, I see a gate there, I need to get a key. How would I get another way to get into this, uh, area, whatever it is. And so it's, it's kind of a puzzle game, but it's this goose that you play as that is, is kind of bad.

It's just like, it's, it is like you, you cause a bunch of trouble. You will turn on sprinklers or chase, uh, animal or farmer, all sorts of stuff. That's the first thing I thought of, but I'm assuming. This, this AI agent's, uh, goose is a little different than that. What is it, what is it designed to do?

Angie: So, it's an interesting story.

Um, we actually built, uh, this AI agent internally for our engineers. Um, we did a functionalization last year [00:03:00] where, uh, we were divided by business units. So we had cash app. They were basically their own company, square, their own company, that kind of thing. And everybody had their own culture and did things, uh, differently.

And so with the Functionalization we brought, uh, you know, different functions together or similar functions together. And so all of engineering is now together, but everybody was doing something different. And um, and so there's a lot of work behind the scenes just to kind of. Streamline things, make things simpler for our, our developers.

We take developer experience very seriously, even internally. And so this tool was built to help developers automate more, move faster and that sort of thing. Um, it did amazingly well internally, and so we decided to open source it. And so it was designed to be like a developer AI agent. So AI agents. If I may, [00:04:00] are, uh, basically just like LLMs you use, like the chat bots you use today, like chat GPT for example.

Well, AI agents will actually do the thing for you versus telling you how to do the thing. And so, um, we designed it for engineering tasks, but then. People internally at our company, were like, wait, we wanna, we wanna move faster too. We wanna be efficient too. And so we have, uh, stretched the limits of Goose and it's no longer considered just a developer agent.

But a general purpose one.

Autumn: What differentiates Goose from what else is out on the market right now?

Angie: One. Um, it's open source and two, it's local. So it's not in the cloud. It's not a website. You download it, you can utilize it as a desktop app or you can utilize it as ACL I. Um, and then [00:05:00] also. It is a MCP client. So for those of you who, uh, may be living under a rock, MCP is the hottest thing out right now in the tech field.

And so, um, that's basically a standard that allows agents to talk to all sorts of data sources. And so, uh, there's probably close to 4,000 of these like MCP service. Servers, which represent like APIs, databases, like there's one for everything, your favorite tools. Um, and because Goose is an MCP client, meaning it adheres to the MCP protocol, it can connect to any and all of those MCP servers.

So you have really exciting times like mixing and matching different things together to do some pretty cool stuff. Um, oh. Go ahead.

Justin: I was looking into MCP the other day, and from [00:06:00] what I understand, it's basically an open API spec based, essentially, it has a couple different fields. It's like, Hey, it's A-J-S-O-N field.

We can get this, this spec from this HDP interface. And all of the MCP servers that I found that were something I wanted to do locally that wasn't an external service were web wrappers around command lines. Like every single one of 'em was like, oh, exec out to FMP. Oh exec out to gi. Oh, exec out to something.

I'm like, these are just AI man pages in a lot of ways. Right? Like it's at the end of the day, how do you use the tool? Where do you get more information about it? And then where do I exec the thing? Is there something else I'm missing?

Angie: Yeah, the part that you're missing. 'cause people look at just the server by itself and they're like, uh, I don't get it.

I don't get, what's the magic? The magic is then that you can. Or these man pages can be used by any AI agent that also speaks that same or adheres [00:07:00] to that protocol. Right? So what that means is like, and, and, and autumn, back to your question, um, I'm gonna tie all this together. Uh, back to your question about like how Goose differs.

So a lot of the, the popular agents out right now are things like claw desktop. Uh, cursor Klein, you may have heard of these. All of these are like IDE focused, right? So I talked about how we are desktop and CLI focused, but even more than that, we, we aren't married to any LLM. So you can bring whatever LLM you want.

You don't have to use, uh, you know, anthropics, LLM or, or open eyes, you know, when they build these agents, it's locked into what. What they're selling. Um, you can bring whatever LLM you want with you, but then you can bring these tools. So Justin, back to your point, like, oh, what's the magic in this? Right? So for example, I plugged in, [00:08:00] um, a, a Google calendar, MCP, and I plugged in an asana MCP.

I didn't have to do anything except say, I wanna use these things. Now I can do stuff. Think about it as the end user and not the Colbys. I can do stuff in natural language now and my AI agent can just execute. So for example, I said, Hey Goose, can you look at my Asana board, find all of my tasks that, uh, are incomplete.

And then I want you to look at my calendar as well. Find open slots, estimate the time and the task to see how long it'll take. And I want you to find places where I can actually do my work. You know? And I don't like a whole lot of context switching, so try to group things together, right? And so I said all of this in natural language and it's able to execute these things because of the simplicity [00:09:00] of the MCP server.

But yes. I like that, that it's man pages for ai. That's true. If you're stuck just looking at the makeup of the MCP, uh, that's probably not that interesting, but what it unlocks for people, that's the magic.

Autumn: I think two things that you said were important. One, the fact that you can do it locally and it's not in a cloud, because I think that's one of the coolest parts about apple's, uh, apple intelligent chip is the fact that it does all of the processing locally.

Turned

Justin: all that off.

Autumn: All the Apple

Justin: intelligence just turned it off. I, it was such a panic. You

Autumn: don't want it, you don't want it. Oh

Justin: man. Every single thing. But I'm have these summaries of garbage I'm going to do.

Autumn: If I am going to use ai, I feel better with the way that Apple set it up. Up. Right. And also what you said with the whole, the fact that you can switch back and forth, it was so cool to see all the MVPs at MVP summit and they used AI in ways that I would've never have thought of, but with the, like some people using T net and some [00:10:00] people using like, you know, um, I don't know, Java or different like.

Frameworks in different languages. Some, um, some of the models were way better at one language than the other. And I think that was one of the cool parts of copilot is that you can lift and shift your model in the way that they've really made it. Where

Justin: we reusing that term. All right. The, the, it's like they were

Autumn: literally just switching back and forth and they knew like, Hey, when I have this question, it is better on Claude about da.

I like that. I

Angie: think that's. Fire. Like I think that's fire because I use it a general purpose, right? So I use it for all sorts of tasks. Sometimes it's coding, sometimes it's more leadership type of things, sometimes, you know what I mean? And so some of the models are exactly right. I feel, I find that, uh, Claude is amazing as coding.

That's who I want code, that's who I want a cold with. My friend, when I went a brainstorm. Yeah. When I wanna brainstorm, I wanna, I wanna talk to GPTI [00:11:00] like Exactly, exactly. I like how they think. I like the reasoning there. You know what, I, results were so

Autumn: much better yesterday on Claude for this very in depth framework that she's never used before.

And I think that's kind of like, I like it when it helps me learn. Like I love asking it questions like about subjects that I don't know, and kind of using it like almost as like a, to simulate what you do in class when you're talking to somebody. Or I like it to do the things that make that are boring, that make my brain just feel like it's, what is the word?

Demand avoidant. You know what I mean? Mm-hmm. So that way I can do all the fun stuff. And I think like we're finally getting to a part where like just pretending like AI is gonna replace us and do everything is just ridiculous. But using AI to make you more efficient so you can do more of what you enjoy and to kind of make your life better.

Those things, I'm so here for. Oh, one more thing. We talked about open source. Can we just talk about how like fire of an open source contributor Miss Angie is Because like she doesn't just do dev reel. Okay. Like she's a speaker, a teacher, and she does like, [00:12:00] like. People will try to talk stuff up to you on like Blue Sky and I'm like, have you seen her GitHub?

Like, like, go sit down. Like you are amazing. Like what was, thank you. Tell us about like all of your open source contributions and just like, or it's anything cool you work. I'm

Angie: actually, yeah. I'm actually a GitHub star, which is very cool. There's less than 100 of those in the world, so it's, uh, a huge honor.

Um, yeah, a lot of the projects, uh, one is. Of course, goose. So that's open source. That's what I'm working on mostly these days. But these MCP servers too. So the cool thing about the MCP servers is that you don't have to be the vendor of the API to make 'em. And so you can make one for anything. Now I'm a automation girly, right?

And so, uh, one of my, my tools that I've used forever was selenium, uh, to do like test automation, browser automation, that sort of thing. [00:13:00] And, uh, my girl didn't have a MCP server out there, so I had to, I had to, I had to do that for the culture, you know, and so I created a MCP server for Selenium, um, playwright shortly thereafter, put theirs out and made a big splash.

'cause that came from Microsoft. And, you know, so it's like. Me against Microsoft. Now I'm joking. I love Microsoft. Um, but, but yeah, like, uh, so those are the kind of projects that I work on. Usually tools that I use myself, I like to try to give back in some way, even if it's not a cold contribution, if that could be, uh, some sort of like community advocacy or anything like that.

I'm, I'm, I'm down for all of that. Nothing's too big or small for me.

Autumn: I think that's really important because it can be such a daunting task to try to learn how to, where you contribute for open source. And I think that's cool. I like to, I

Angie: like to go to the docs. People need doc help all the time. Hundred percent time.

Yes. And people love that. Like if you're trying to [00:14:00] contribute to a project and they don't know you and you're trying to contribute to their code base, everybody's kind of looking like. Like who, who this, you know?

Autumn: Yeah.

Angie: Um, but docs, people are like, oh, yes, by all means, please. Thank you. You're afraid at that point.

There's always something. There's

Autumn: always something that people forget when setting stuff up, and they've never, because they've used it so much before they release it, that if you set something up, there's always a good tidbit that you can add. Yeah,

Justin: always recommend start if you're gonna contribute to a new project, especially in like a new community, start with docs because you'll understand how the process of the project is, like how many

Angie: That's true.

Checkoffs do

Justin: you need, right? Is this worth your time? Like start with your effort really low. Just say, Hey, I fixed this in the docs. And if they say no, you're just like, okay, well maybe I don't want to contribute here anymore. But if they're like, oh, you know, can you add this? Here's some corrections, whatever.

Oh, could you also do it in this other area? Oh, they need this help. They want it. You know? Yeah. And that sort of engagement's good. But also if you're contributing docs, please, please do not submit AI reviews of, like I've had, we've had poll requests and people like, I rewrote [00:15:00] this doc with ai and I'm like, please, no.

Like this is not an improvement for us. Like we want people to spend their time 'cause. The, the limitation of open source is usually the maintainer time, right? They, they just can't, you can't take all the inbound requests. You don't, can't do all the software, but all those reviews just take more and more time.

So if you are just throwing stuff out there to get your name, please don't.

Autumn: I don't know why people don't like use that as their first pass and then go and like look over it. 'cause I'm like, would you trust, like we'd look over human stuff like make sure you code review your AI stuff. But yeah.

Justin: Can you describe like what was your process for writing the MCP surfer there for.

Uh, not for, for Selenium F Seum. Yeah.

Angie: Oh, look at you. What, what, what did

Justin: you do? Like, how do you local test that? How do you make sure that it's gonna work properly? There's actually,

Angie: yeah, yeah, yeah. Okay. Good question. Oh, I can geek out about CP all day. Um, okay. So there's actually official SDKs and uh, type script.[00:16:00]

Java, Kotlin, Python, rust, and Microsoft just released a C one like two weeks ago. Um, so you can pick up. The official SDK and everything's kind of in there already, so it's really straightforward to use. You basically just initialize a server and then you start adding tools to it. So, um, there's like a server tool and then, you know, just like you said, you kind of put the name, the description, and then like the, the body of the function of what does this do when this tool is called.

And then there's a lot of community members, well not a lot, but a few community members who. Are simplifying that SDK even more by making what they're calling fast MCP. So as far as I know, there's one in Python and there's one in TypeScript, but with the fast MCP, it's supposed to be even simpler. But I didn't use Fast MCPI wanted to use like the og, SDK, you know, the og.

It's been around for three months. Yeah. Yeah. Yeah, yeah, yeah. [00:17:00] We move fast over here. Just, um, but yeah, I, I used the, the OG type script, SDK, which is interesting 'cause I'm a Java girl, but as I was writing it, the Java one wasn't out yet, so I was like, all right, I guess I'll use type script. And, um, actually.

I use Goose quite a bit on, uh, like, okay, what am I missing? What functions should I add? Is this cool? Is this not? And so, uh, it basically helped me in writing it. And then there's actually a MCP. Specter. That also comes from model context protocol, and that's how you test. So you can test this locally by just kind of hooking it up to this inspector, and it'll show all the tools that are available in that MCP server, and then you can try to execute them too and make sure everything works before you deploy it.

So then once I made sure it works, commit that to GitHub, make a NPM package [00:18:00] and good to go.

Autumn: Usually way better than Comcast. Angie, I know for a little while when you at least started at Block and and TBD, um, you were focusing a lot on identity. Where has that work kind of gone? What were you doing with that and how was that kind of playing out? Is that shifting into this sort of AI space or is that something that you're, you're kinda like, Hey, we have a standard and we want people to use it.

Angie: Yeah, that was totally separate. So I lead Dere for all of block, and so that was one of the projects. And so that one was, uh, yeah, digital identity and cross border money movement. And I was under the, um, umbrella of TBD. That business unit has such, has since shut down and we have donated those projects.

All the open source work that we did, we've donated it to the, uh, decentralized Identity Foundation. And so, uh, they're, they're maintaining those projects now. There's a new foundation I didn't know about every day. Mm-hmm. That's really good that you [00:19:00] all donated. Donated it all. Yeah. We don't want the work to die.

Um, it, it was, it was good work, you know. Um. As companies reprioritize, you know, we just kind of needed to focus elsewhere. But yeah, there was a good momentum behind it and everything like that, so we didn't wanna just kill it and we wanna be good open source stewards and so, uh, donated it to make sure that it stays, stays alive.

Justin: Is this the independent foundation? I don't see any, like, it's not Linux Foundation it looks like. Right, right, right. Yeah, they're independent. Okay. Uh, that's an, I'm trying to see, I'm looking on their website, like what other. Projects or things they've done in the past, because I know some foundations are, are different than others, right?

Like some of them are very focused on an industry, uh, and then some of them are kind of where projects go to die, right? Mm-hmm. The Apache Foundation has a bunch of projects that are kind of like the geriatric home for open source. Dang, Justin. That's why people ask you if you have hot takes and webinars.

Autumn: This is exactly [00:20:00] why I, I stole that from someone else, but I liked it because it was, it's, it's true that there are a lot of foundations out there that are just like, Hey, we, we want this to exist in the world, but we ain't spending any more time on it. Right? Like, someone else needs to maintain this for us going forward.

Justin: And, and kinda the community's gonna shift off at some points. Um. I, I don't, again, I don't know anything about the DIF foundation or what the projects are, so I'm just kind of curious seeing, there's too many foundations, there's too many places that the software's gonna land and there's too much code that's not being maintained in all of them.

And I kind of wonder how that also is gonna play out. Yeah, well, diff is, uh, has been a great partner. We worked with them a lot as we were developing things in, uh, the decentralized web node stuff that we developed actually was their spec. And so it felt like a good home. You know, they created the standard for it.

Angie: They, uh, contribute to the code. They promoted very heavily while we were developing it. And so, you know, they, they were more than willing to, [00:21:00] to accept. So your implementation was implementing their spec? Yes. Oh, cool. Is it exciting being kind of the head of like Cash app and the all the other different, um, kind of technologies that block has to get kind of like the variety of different aspects of Block?

Uh, yeah. So Head of Devereux, let's correct that. I wanna, anybody think I run all of cash out? Yeah. Yeah. Uh, but yeah, the, it is exciting because I'm tech girl, right? That's my, that's what I go by is like my username across the internet is tech girl. And I don't ever wanna box myself into like one specific tech.

There's things that I go really deep on and that I feel I'm an expert in, like. Java, for example, you know, things like that. But then, um, I also like to learn and dabble with all sorts of things and be pretty knowledgeable, at least at a surface level about most of the things, you know. And [00:22:00] so, yeah, this gives me a, a great opportunity to, one, explore these different spaces, but two, look at like how you can connect them.

So one of the things about me is I'm an inventor, so I have like. 20 something patents. Right? And so I love to kind of think of creative ways to utilize tech, to merge like things together that don't necessarily belong together and see like what happens. That's so cool that you have that many patents.

Autumn: Thank you. You are amazing. I wanna be like you when I grow up. How did you become a Java champion? Oh my God. So I love Java in case anybody didn't know that. Like, I love Java. So I, I actually started, uh, coding in Java in college. I started with c plus plus actually. And then, um, I had a professor at Java, professor, like maybe my junior year, and he was a, a young.[00:23:00]

Angie: Uh, Java programmer. I think this might have been his first teaching gig, but he was just so on fire about Java and that just became contagious and the way he taught it just made it very. Enjoyable. And so, um, he, he was so nice that after like the first semester he did, maybe two semesters, he did a, a bootcamp over the summer so you could go home if you wanted to or you could stay back for the summer.

And he did a, a. Free bootcamp for students who wanted to become certified in Java. And so I stayed for the bootcamp and, and we went hard and we learned the ins and outs of the language and, and all of this stuff. And, uh, we, I took the test and I became certified Java programmer before I even graduated college.

Which was very marketable to get my first job. Um, because back then Java was the hot new thing. That was the hot new language. So the industry folks didn't really know Java. This was something he was just trying to pick up. And here you got like this girl coming outta school, what, [00:24:00] what a certification is.

So that, that was very, um. Very beneficial for me. So anyway, I did Java for pretty much all of my career, um, and not just did it, but also shared the things that I was doing with, by like teaching courses and workshops. Uh, I founded Test Automation University, which is absolutely free, and it's uh, just courses that are available that teach people.

Not just Java programming, but different programming skills. How to automate, how to test, how to do all of this stuff. Um, and so with all of those contributions, um, you are recognized in the, the Java community. And it's in a, it's a, it's a distinguishment that, you know, has to be bestowed upon you. You don't get to apply or anything like that, but, and it has to, to be voted on.

It has to be voted on, and it has to be, um. You have to be recommended by someone who's [00:25:00] already a Java champion. At the time, there were less than 300 of those in the world, and so it is a very small community of people, but they have to know you in your work to be able to put you up. And then, yes, the rest of the Java champions.

Would vote like, do I know this person is there? Is their work actually impactful? Um, and fortunately everybody say yes. And so, uh, that's how I became a Java champion. It is a very hard clip to get into. Are you still the only black women woman Java champion ever? Yes. When you gonna join me? Autumn? Look, I'm trying, but I just like, I thought that way you'd, when I was in college, everybody that was an engineer did not look like me.

Autumn: Nobody in the, like most women were in like front end. And then here comes Angie Jones, the only black female Java champion and I love Java. And you were just like. The biggest star ever. Like I was just like, it, she has cool glasses and I wanna be just like, [00:26:00] like you were like the first time I could ever like see myself, but also as like a backend programmer and everything.

Like, I just thought you were the coolest thing since slice bread. Like I, I so appreciate that. Um, and I'll tell you why I appreciate it. One, it's just really sweet, but I also appreciate it because, um, I, I definitely want to break down barriers and doors. And help people, help other people like me see what they can be like.

Angie: When I was coming up, I didn't know anything about programming. Like I, that just wasn't on my radar at all. Like nobody I knew in the world. Um, was a coder or anything like that. So if I didn't see it, it's just not even a possibility. Like I literally did not know it existed. Right. I, I stumbled into a programming course, not even knowing what it is.

I just thought, oh, my dad told me, actually, you need to go learn about computers because they're about to take over the world. Kind of like this AI thing. Um, but back then it was computers and so I, me [00:27:00] not knowing what. Uh, computers do or whatever. I just like found one course that said something about computers and that was a c plus plus programming course.

It wasn't like an intro to like windows or something. It was like hardcore thrown into the deep end, and I actually loved it. So that's how I looked up and got into it. But if not for that. I'm, who knows where I would be and what I would be doing. And so it's really important to me to be, you know, representation for other people like me so that they know that it exists, even if they choose not to go that route to just know that it exists.

But what's really special is I don't. I'm not a person that just walks around talking about that sort of thing. No, you are, are so humble. I'd be wanting to come for people on the internet. They'd be like, trying to say stuff about you and I'm like, I will tell you her resume, just because she is humble.

Autumn: I'll do it for her. Like, like You're amazing. Yeah. But I, and, and I don't want, I don't, I don't want to. I don't like I'm [00:28:00] in this space 'cause I really love tech, so I don't want to even be distracted and talking about things like being a woman. Yeah. Or being black. Like I don't want you to like, I don't want that to be what the conversation is about.

Angie: Like I'm, because it really does get to be it about AI and mcps. Like let's talk about that. And so I really appreciate like when you're like. Oh, I like you. Not because like, oh, you're a black woman because you're doing dope stuff, you know, in tech. Like, I like that. Well, I, I remember that like, so I love computers the whole time when I was younger and then.

Autumn: I took my first web design class and I hated it. I was like, this is not for me. Dreamer sucks. Like I don't ever wanna do this again. And it was so funny 'cause like I watched Hidden Figures and I was like, I love computers so much, but I don't know why. Like I all the like intro classes they put me in.

Absolutely hated it. And then I got my first Java class and I was like, oh my God, I love this. And it was crazy because nobody. Like everything that was ever like computers, I didn't know that it existed, [00:29:00] like being a computer programmer existed. But anything that was like, I guess shopped to me as like, this is what girls do in computers, was never the things I liked, you know?

Mm. So seeing like you, and like the way that you talked about Java and you were so excited and everybody was always like, I'll try Python, or I'll try, I don't know, JavaScript, but Java's too hard or it's too complicated. And I was like, I love the like. Just the structure and the fact that it is complicated, you know?

And then you were the only person who was just like, you were so excited about it and you were just like, and then I built this thing and then I built this and that, like added it all together. And I was like, finally, like someone like me, but like, even cooler, like, I wanna be like you. So it was just so cool.

Like you, you didn't shy away from any of the hard things. You know, like you never, and it's so hard because sometimes people will be like, come and do this talk because you're a black woman. And I'm like, can I do it? 'cause this thing is really cool, you know? But like, yeah. And, and I'm dope and I know my stuff.

Yeah. Like, but then also like, I do really appreciate that I [00:30:00] do see myself in you, but it's not like you're only skill, you know what I mean? Like you're not only known for diversity. And I'm like, I try so hard to figure out like how you did that, because I wanna do that. 'cause I don't wanna just get like.

Given diversity talks, but it is important to be in those rooms and to speak up for those things. But like you are just such an amazing contributor to so many projects and you do so many things, and like, I just love that it's never like, you don't ever shy away from the new thing or the hard thing and you like, it's almost like a great light, an example to show us how to do it, you know?

So I just appreciate you. Thank you. You've, you've switched a lot in the things that you focused on over your career so far. How do you. Choose what to focus on. Right? Because there's so many things, right? You, you have to say no to a lot of things. Once you're at a certain level of like, I understand what's possible.

Justin: Now I have to start focusing and say, I'm gonna invest some time here instead of over at the other thing. Right? Like, how do you make that decision? I. Yeah. Um, I like to be on the cutting edge of tech. Um, that just keeps everything really interesting for me. [00:31:00] Um, one for myself, but also one gift that I have is being able to explain concepts like very simply and help people understand.

Angie: And so I like to get ahead of it before. Four is the main thing. Like the, like the whole MCP thing. I was there on day one in November when it dropped and I was like, oh, yo, this is so, like, this is dope. Let me learn as much as I can about it because it could take off. I was hoping it would, but I wanna be able to explain it to people, you know, and help them figure out like how to utilize it in whatever they're doing or how to, uh.

Avoid certain problems. I wanna know the ugly, the bad, you know, the good, all about it. Same thing. Um, when I went, when I worked on the stuff with TBD, and so at that time everybody was talking about decentralization, but people were going the wrong route. You know, they were like all into crypto and, oh, we need crypto for [00:32:00] everything.

Like, we shouldn't be able to even post on Twitter without. Put some Bitcoin on it and it's like, yo, okay, I don't think that's the right way to go. Let's find like some other ways to explore decentralization, maybe using like smarter technology and things like that. So that's always kind of been my approach, like what's what, where is the industry headed?

Um, people always ask me, what's your long-term plan? I don't have long-term plans. I keep myself very much so open and available so that I can easily pivot when I need to. You know what I mean? And so right now I've actually been talking about AI Before generative AI was, you know, a hot thing before it was.

Accessible. Like I could see, all right, this is coming pretty soon. I didn't know it would take this form, but I knew it was coming pretty soon, and so I studied that way back. You would, if you look me up from years ago, I have talks and things on ai. I was already working in AI in some form or [00:33:00] another, and so now that that's landed, I already have that expertise.

I can come in and kind of help, you know, others who need to adopt. I also appreciate that you, you'll find a new tool and you don't like car salesman. It, you know, like some people will just be like, it's new and then you should use it for everything, and we're all like. But should you, you know, but you'll be like, Hey, this new tool is good for this and you can do these.

Autumn: Don't try to do it on everything. That's right. Yes. Like you can do it and use it for this tool and it will help you to do this. Like, you have very good use cases and very clear use cases and it never comes off disingenuous. Like I feel like I just side eyed all of, um, decentralization until you got into it.

And then I was like, okay, maybe 'cause Miss Angie's talking about it. You know what I. I feel like you, when you co-sign something, you have very, um, real [00:34:00] ways to use it. And you don't just give ev like, you don't just put your name on everything. And I think that even goes back to what we were talking earlier with like how honesty is expensive, but it also like, it makes us trust you.

Like when you put your name on something, it's completely different than just any person in dev reel, you know? Yeah. And that actually leads me to one of the questions I wanna ask. When should I not use MCP? Right. Like when, when is it not a good fit? Okay, so, um, there's a lot, I'm not gonna say not use MCP, but like use it wisely.

Angie: So there's a lot of talk right now about the security of MCP. We're very much so yo lowing this right now. It's so new. Um, and there's a lot that can be matured, um, in the space. We're actually using MCP very heavily internally within blog, and I wanna write a blog post. I just gotta find the time to do it, but I wanna talk about like.

For enterprise type use [00:35:00] cases. 'cause this looks very different than the vibe coding you do on Saturdays when you're just like, oh yeah, here's, here's five mcps I'm gonna throw in here. You know, from, from my boy I never heard of on GitHub. But uh, that's so real though. People just be using anything and I'm like, you can be using anything.

And I'm just like, that's. That's different when you're like, oh, I wanna build this really cool thing, but like the security person to me, my, I be twitching. And they're like, and now I built it and connected it to all of my data. And I'm like, oh, why? Yeah. That's nuts, right? So, uh, so, so there's like a lot, a lot of ways I'm like looking at, okay, how do you use it responsibly When, like, how do you suss out different CPS internally?

We, we basically have, um. We create a lot of our own, even if one already exists, we'll just make it ourselves. And so that way we know that it's safe, um, and it's in-house. [00:36:00] And we also help, like I told you, we not just, the engineers are using it. Everybody wants to use it. So you got people in, uh, HR in, uh.

The finance department, the, the executive assistants, the marketers, like everybody is using it and they wanna connect to these systems that I never even heard of. Like, these are systems that I don't use, you know, like these are for the finance folks or something, you know? Um, and so they're like, oh, I need a, I need a MCP server for this.

And we are like, oh, okay. But all of this is being connected to very sensitive data. So we create those CP service ourselves and keep them in-house, but we also. Uh, do a lot of the plumbing for them, like they don't wanna be. So for MCP server, anybody who hasn't done it, you're like copying JSON and like putting it in a file.

Like it is real, it's real raw out here right now, you know, like you're copying like MPM commands and stuff like that. So. You know, [00:37:00] Ted and finance don't wanna do that, right? And so you're like, okay, let me, let me hook this up for you. Let me like wrap it and o off. I'm gonna hit, hook it up real nice. All you gotta do is press a button, you know?

And so. It's pretty sweet. That's what I'm excited about. Like I understand that AI is where we're going, but I want to really help educate people on when to use AI and how to make it safer. Because, because we're giving people the ability to use technologies that they may like be very new to them. I think it's gonna be really important to help educate people on what those technologies do and when you need to be like more careful with using them.

Yeah. So let me tell you this. Um, and this is a real thing that's, that's happening. My sister is not in tech and she, she text me the other night and she's like, yo, you all around the world teaching everybody else about ai? What, what about me? I need to know it too. She, because she's, she's very aware that it's [00:38:00] becoming so mainstream that you're going to have to use it.

It's like, it's like. Trying to apply for a job and you don't know like how to use, it's like the cell phone before, like not having an iPhone or not having a cell phone People act, but go bigger than that because that's more personal. Like this is work. This is like. People are going to have to know how to use this.

Like, oh, I don't know how to type on the computer. I still just use the typewriter. It's like, all right, you're not gonna get the job. So she recognizes the shift that's occurring and she wants to educate herself, but she has no idea where to even start. Like it is so much noise out there right now.

Everybody's trying to sell you something as the end all be all solution like, and she doesn't even know where to begin. Two days later, my aunt called me, aunt Connie called and said, baby, you got some AI classes I can take? And she's like, you need to, you need to make us some classes and, and give me a little certification or something to go with it.

That's Aunt Connie. So people, I [00:39:00] love that she knew you were the person that makes the classes certifications, so that's fire. She was like, she'll make one for me. It'll be fun. But, but everyday folks, this is not just a tech thing. Everyday folks are realizing like, this is something that they need to learn, but what are we doing?

Uh, to help people get there. Like there's big pushes in companies too that say y'all need to use AI to be more efficient and to move faster. Sounds great on paper. How do I do that? And to your point, how do I do it safely, right? And how do I do it efficiently? And I think that we, like, there's so many aspects we need to teach, you know, aunt Connie, how to do it and how to be safe.

Autumn: But also there's so many older people in our like lives that like we have to teach them how to be safe from ai, not to give too much of your information, where to use it, but also there's, we're pitching it to startups in a way where. Like, and I mean startups that are not just tech startups, right? Like people that are starting their own business or we're saying like, you can get rid of your HR department, you can get rid of this and just put this AI in it, but [00:40:00] you're putting personal, full, personable, identifiable information into these things, right?

Like we need to, like that's cool that you wanna use it, but are we using it in a way that's safe? And do you still have at least one human that knows the regulations in that industry that you're doing to make sure you're safeguarding your business towards that? You know, autumn, let me tell you this.

Angie: Lemme tell you how, how quickly you can get caught up. So I had a deadline last night from the IRS 'cause they were like, oh, you owe us money from 2023. And I was like, I don't owe them no money. And it had like all of my stock stuff on there from my company, which, you know, your company pays that already.

So I had to, I had the letter, same problem. Yep. So I don't know that kind of stuff that well, like, I don't know, IRS speak that well, you know, and so I'm, I'm with, I'm with Chad GPT and I'm like, help me craft this. This is what's going on. And it was so tempting, even for somebody like me who knows better, it [00:41:00] was so tempting, like, damn, I should just upload my, my tax returns in here so they can tell me exactly.

Where the discrepancy is, but I knew better not to do that. But somebody else, like Aunt Connie wouldn't know that. She would just put it right in there. Every day in regular day life, I'm like, oh, this is a task that my brain does not wanna write this thing. And I'm just like, oh, but can you just imagine how this is gonna go on Twitter security expert puts her like, you know, you're just like, oh, you know better, but it won't be so easy.

Autumn: That's okay. You work at Microsoft, so you have an out. Justin. Okay. The best part of this whole podcast is Miss Angie's faces that she makes every time Justin say something ridiculous. I was throwing, like, throwing whatever out there. Also, you look so pretty today. Thank you. Thank you. Oh, I wasn't talking to you.

Angie: You do too. Thank you. My, my last question here. You mentioned when you good to Justin. Oh, [00:42:00] when you, he's like, finally, someone's being nice to me. I'm trying to ask a question here. I'm trying to be serious. Keep us on track a little bit. Uh, when you said you, you graduating college, you got your certificates, it helped you kind of get a leg up in the industry.

Justin: Uh, your, your aunt or whoever's calling you saying, I need to get certificate or some classes to up my skill level in this area. I talk to a lot of people that are new in tech or trying to get in tech. And I don't think that. In today's world, especially with AI as a tool chain, we can rely on for a lot of things.

I don't know that certificates are the right place to point people to say, you're gonna get recognized or you're gonna get a a some advantage if you have this cer. Mm-hmm. Certificate rather than person that goes and builds something or, or tries something and maybe even fails at it, but then publicly talks about it.

Right. Where do you think people are falling in this, in today's tech world of, I'm trying to use technology in my job. I'm not a technologist, or I'm not, it's not software as the thing that I do, or people that are trying to get into [00:43:00] software and say. Uh, do I get all the AWS certificates to be a cloud engineer?

Do I go get the Java thing to be a Java developer? Where do you think people are trying to educate themselves and where are you seeing that businesses are rewarding them for their efforts to give them jobs? I'm gonna say something controvert. I didn't want, I hate to go here, but I gotta tell y'all the truth.

Angie: So for some people it's different. Justin, you need, you need some paperwork behind whatever else you've done. You could have built the things. So in college I did three internships. I did uh, two with Boeing, and I did one with IBM. People wouldn't talk to me until I had that Java certification. So even though I had exp experience, I had more experience than than any college student you're gonna talk to.

I spent like, not just the summers, I spent a semester doing college work and working at Boeing at the same time. Like I had the, the receipts but it, without that [00:44:00] certification, I couldn't break into tech. You know what I mean? So for some of us, we don't get like that benefit of the doubt, like, your work is enough.

There is no benefit of the doubt. You gotta do like five times more than everyone else. Yeah. And they still think you're some like. DEI hire or something. Yeah. And so and so, it's like, oh, okay, you took the, you took a test that I know, I know that test. Right. Um, I know how hard that test is and, oh, okay. You passed that.

All right. I guess I'll give you a shot. I also think that. Certifications don't, I wouldn't say they teach me things, but they give me a deadline in which I have to learn things by, right? So I can't study for a test in the same traditional way. Like it's not like a test isn't going to make me know things.

Autumn: And I don't think it's necessarily the best way of to it's building knowledge. Like I don't think you can prove that you can build cool things with a certification, but it will. It will trick my brain into really focusing on a subject and diving deep in it [00:45:00] because I'm like, oh crap, I have this certification that is coming up on the first, so I need to know all this stuff.

So the motivation side of it. Yes. Yeah, that's a good point. Conference driven to develop. No, that's a really good point. And it gives you some sort of success even if you don't break in. So I have a cousin who's studying. Uh, cybersecurity. He actually works in a health field, but he wants to switch to tech, and so being able to just obtain these certification gives him like, oh, I'm doing it.

Angie: You know, I'm actually doing it. Even, even if he doesn't get a job just yet or whatever. It might also help that pivot, like, maybe you don't get a whole cyber security job helps, but you get a more in healthcare cybersecurity adjacent job. Yeah, yeah, yeah. Something like that. So I don't know. And I don't know every space, like I feel like security certifications are important from what I see.

Government type gov tech type thing. They're definitely industries that value them more than others. Also, honestly. Yeah, they have sec, they, my cousin is transitioning back into work after being a stay at home mom. They've made her get a [00:46:00] customer service. Certificate, like people agon certificates for everything.

Autumn: So like honestly, if like it, I think it depends on the level of job. Maybe if you're just starting out, that's a good realm. You know, like if you can find, especially like a lot of libraries will have ways to get your certifications for free or like different things. Like I wouldn't go. Like spending your life savings on it if you don't have to.

But like, I would definitely look into ways that get free or low cost certifications. 'cause like, I mean I've taken actual like classes from Angie's like test automation University and I learned a lot, you know what I mean? And that's a free resource. So I think there's, there's ways like you can audit Coursera classes and then if you really want it, like get like a certification from that university.

So I do think, I think sometimes like. Some people just need a way to break in, but I don't think, like, I think it's different when you work for a major company and they force you to get cert six certifications, like you're gonna all of a sudden be a master at that. And I'm just like, ugh. Because we've all been there like, you know.

I have a [00:47:00] one last question. How do you think cost will play into all this? Because right now a lot of this technology is be giving away for free, right? So like, do you, like, how do you think that, like when we start having to charge for the, like the amount of compute power that goes into all these things, like do you think that like.

Like, you know, an API versus, I don't know, an agent. Like, are agents going to end up being like way more expensive? You know what I mean? Like, I just wonder how cost will play into all of this, because that's a great question and I am, I'm already seeing it actually. So for example, I, I told you all Goose is open source is free to use.

Angie: It's local. We don't track anything. It's whatever. But you gotta bring your own LLM, right? And so if you bring, uh, one of the ones, one of the powerful ones, you have to pay for that. And I'm like, at first, when I first started looking into like L LMS and stuff, I'm like. Wait is stuff not [00:48:00] free? Like you're just used to everything being free.

Like what? And so it's not free and I've been actually observing this, uh, and it's interesting to me that developers who are like the cheapest people on earth, God, Jesus, are willing, are willing to pay for LLMs usage. Like they pay, I see them pay like a hundred dollars a month and stuff like that. It's like.

Interesting. But then I started thinking about like, what does this mean for junior folks or folks in, you know, areas that are not as rich as like America, whatever, where you don't have an extra a hundred dollars to blow on like, you know, LLM usage and like playing around with stuff. So it's, it is really interesting.

I don't have any answers. It's just things that I've been observing over the last couple of months. We're definitely in the Uber VC funding phase of this right, where we can offset all of the promise of, you'll make money in the future by saying, [00:49:00] Hey, those rides are cheap right now, but as soon as, as soon as the bill, soon as the taxi system adjusts, or we have some other regulation in place, or we have to actually pay our drivers better or give them any benefits, like all those regulations cycle up.

Justin: Oh, that's actually, I was hoping it was gonna go down, Justin. I was hoping it was gonna, I was hoping this all just becomes free, like with the hopes of like the deep seek and then you need to compete with deep seek. So you gotta give me a really good model, open source, and free too. Well, I mean, the, the freedom of it can cost you, you know, your time.

Sure. If you're gonna run it yourself, if you have the expertise and you invest the time to learn how to do it yourself. Like I have GPUs right behind me. I run a, you know, LLM models on it. Like we can do that. At home, but I had to learn the skillset to do that. And not everyone can do that either, right?

Mm-hmm. There is that overlap. That's true. Gonna fight Microsoft. You have an extra hundred bucks patching, you can go. A hundred extra a hundred bucks, you can go pay for it somewhere. If you've got an extra five hours, you can go learn to do it yourself. Right? Yeah, that's true. True. But somewhere in there true.

You have to be able to invest in it [00:50:00] somehow. And, and a lot of people coming into the space, they have a lot more time than money so they can try to invest in it. And that's how I got into technology was I was like, I have all the free time in the world and $0, I'm gonna go learn Linux. Right? 'cause I didn't have money to buy a Mac or use Windows.

I couldn't pay for it. So I'm just learning this other thing. Yeah. Right. Oh, wow. That. And so I do think that that's, we're at, that's really interesting. Yeah. But I don't, yeah, I mean, there, unless, unless there's a subsidiary. At the government level, right? Like where we have to keep the cost down, right? Like if I look at my electric bill, right?

Like I can pay for electricity. There's only one company that gives it to me in my area, but they have, they can't gouge me, right? Like I'm a, a captive audience. They, if, if, if there was no regulation on their industry, they could charge me as much money as I would pay to have power at my house. Yeah. But there are requirements that they offer this at a reasonable fee.

So unless we're doing something that offers. At a subsidy government sort of law that we say everyone needs access to this tool because that'll level up [00:51:00] everyone. Which I do think, I mean, could be a good, right, a good option there, but then someone has to run it. Right. This isn't free. Yeah. This does cost a lot of money.

This is a lot of compute. Yeah. And, and maybe. The cloud providers say, yeah, absolutely. We're gonna go into that program. We're not gonna make a much money from it. But it might get us some other fringe benefits around it. But that's also how people start going back to on-prem from the cloud. Right? Because before, before when they were giving it away for free or giving it away for cheap, yeah, all those were credits.

Everybody used it. So I think the more that they start charging for ai, people are gonna be like, do I really want this agent or do I just wanna go back to APIs? So I think it's going to be interesting to see how that shakes out when you have to pay for all these things. Because at some point, like. They open AI is not gonna keep raising billions of dollars.

Autumn: You know what I mean? Like, no, but the, the point is that this stuff is not free right now. Like I use AI extensively at work. Mm-hmm. Because I'm on Jack Dorsey's bill. Yeah. When I'm at home and I'm working on my side projects, guess what? I ain't [00:52:00] cranking it out like that. I'm, I'm using, I'm using the old school.

Angie: I'm back on the typewriter. But see, but see, here's, but here's the, but this is what the, that's actually a really good point because the difference is, is Miss Angie has that skill. And we are creating a whole generation of people who will not have that skill because they will only come into the world with that.

Autumn: We are selling it to people as we're telling them they don't need to learn it. Like, did you see, you don't go to school? Like they had more Cuban and all these people. That's kind a good idea. Exactly. Like he was like, you could be like almost a doctor if you just like have chat GPT, you can learn anything.

And I was like, no, bro. Like you have to actually learn things and then you use it to like make you more efficient and to learn those things. But it's completely like, that's why I think you using AI is going to be like so cool because you're just making yourself more efficient and learning faster. But you also know that first programming language, just when you don't even have ai, knowing the first programming language, knowing the second are two different things.

Right? So can you imagine you have years of deep [00:53:00] open source contributor and like. Engineer and just all like you and teaching people how to do things is such a deeper level of like knowledge when you've like learned how to teach people. So like I think if we, I just hope that we don't lose that because people are going to need that, you know?

Angie: Yeah. We're, we're really playing with fire right here. Yes. I think right now I really do in, in a lot of different ways, but mostly like that accessibility Yes. To this, to this technology. Um, and the, and the thing about, well. Um, one, you need to be using it even though you might not have access. And two, forget trying to learn how to do it on your own.

That's played out. Mm-hmm. Like both of those are very dangerous. Yeah. Mm-hmm. I think that as you, oh. Go ahead. I was gonna close. Oh yeah. I just think that like, it's so interesting, like companies are pushing it, but they're ruining their future workforce because they're going to want people who are curious, who are trying new things, but they're also cutting that off at the knees.

Autumn: So I just wonder where it will all [00:54:00] go. And plus, like I said, we're not hopping as much anymore, so you have to now be there to figure out the results of your bad decisions three years later. I mean, I'm just gonna say that's what they do. Yeah. Microsoft. It's not the first time they made a mistake. Steak, you're not wrong.

Justin: Microsoft's 50th anniversary party. Satya said that maybe an AI will run Microsoft in 50 years. That's what he said, right? Like this is, that's what he said. He's like, this is, maybe we don't have another CEO after him. Maybe it's an ai. All right. Whatever. Like you go see if that, how that works for you. So anyway, mark Cuban literally said, you don't need college degrees or to learn things.

Autumn: We'll, just, you can just use ai. And I was like, I don't think you can just walk into any, like, I think we're almost like not appreciating what humans do and the reasoning and the context. Mm-hmm. You know what I mean? Like. Just the fact, ain't nothing artificial about making some mistakes in the real world, right?

Justin: It's so important. Like that's part of being an engineer is really being bad at stuff for a while. Like Angie, thank you so much for joining us [00:55:00] on the show. Where should people find you? Angie jones.tech. And so I have all of my links there for my social sites as well as all of the upcoming conferences that I'm gonna be at.

Angie: So definitely check that out. Thank you so much for coming on the show, like you've been like my heroes for like forever, so I'm so glad to finally get to Thanks for having me. Yeah, and thank you everyone for listening. We will talk to you again soon.