Advancing Open Source and AI: Insights from Intuit’s Jimil Patel & Lisa-Marie Namphy on the Argo Project

Hi everyone. We're back here live at Cube Con. It is, uh, day two.

Well, it depends how you count. 'cause cube con's funny. 'cause we have, like this day zero all of the, uh, satellite conferences, but we don't really, that's a zero.

Yesterday was day one today, therefore's day two, though. Some of us have been here three days. Yes.

It's fuzzy meth. Anyway, though, we're, we're still live here. My next two guests are with Intuit, and I know what you're saying.

Intuit Cloud Native Computing Foundation. What are they doing here? Well, I'm gonna let them explain it, but there's a really good story behind it.

Let me introduce you to them. We have Lisa, Lisa Marie, not Presley, Lisa Marie, Nancy, who, if you've watched our previous, uh, Textron cover, uh, CubeCon Coverages, it's not her first time here. And then we have Mr.

Patel, and I'm blanking your first name. Jamil Jamil Patel. Also from Intuit here.

And, and guys, welcome. Thank you so much. It's great to be back.

It's always good to see You act. Always good to see you as well. So I, I, I kind of opened the door with the question into it.

What are you doing here? Yeah. Well, that first day of the conversation, you talk about nothing, zero about it.

We, uh, we actually started, uh, one of the colos. It's the Argo Con. Yeah.

'cause as you know, Intuit is the creator of the Argo Project and donated it to the CNCF started that conference, um, with the, with tons of help from, um, the Acuity and Codefresh and Red Hat folks. Uh, and now it's one of the most popular co-located events. Octopus Octopus deploy now.

Yes. Yeah. Uh, that's true.

That's true. Um, so, yeah. And, and those, we are all still incredibly active in, in the Argo community.

And Argo Khan is one of the most popular co-located events. It was packed on the first time. Well, Argo CD, I believe is now the number three, uh, project in CNCF, right behind Cube UB itself, and then hotel over here.

Yeah, yeah. Yeah. And there was a stat at Argo Con, uh, two thirds of all Kubernetes low, uh, users have Argo CD installed.

Really? They just republished, actually, as of yesterday, Dan just talked about it. 97% is are using Argo.

It was a crazy stat, which almost says that more people are using Argo than Kubernetes. So that math we need to sort out, but people are using Argo. People love Argo, particularly Argo cd.

Yeah. Um, so yeah, no, it's great. It was very popular.

Uh, but, but to your question, open source is we're, we're on the platform team. Yes. And open source is pretty core to a lot of the work we're doing.

And we don't just, you know, consume open source. We also do build and contribute back things like the Argo Project, the Raj, um, which is our Kubernetes native event stream platform. And all of this stuff we use, we consume, I would say more than what, a hundred open source projects.

Mm-hmm. Or contribute to. So we know we, we've built a little open source program office.

We'd show up at events like this. And I just had a conversation with someone who walks through our booth about the hundredth conversation I've had, where they're like, you know, what is Intuit doing here? I I didn't think of Intuit as a technology company, and I definitely didn't understand how important you were to open source and vice versa.

And we've won End User of the Year award twice from the ECF really? Here at Cape Cod twice. Very cool.

Um, so yeah, we're very proud of the work we do in open source. So I of course, knew this before we started the conversation today. Um, but when you think about it, you know, one of the things that makes the CNCF successful is the mix.

It's not all technology vendors, it's not all, and I don't mean this in a bad way, geeky, you know, uh, dev or ops or, you know, A lot of end users Verticals. But it's also end user organizations that quite frankly have the resources to, to dedicate people and time and money to open source projects. And instead of kind of keeping it for its themselves, which is kind of the old way of doing it, become good community members here because they're vital to the mix.

Right. At the end of the day, you are not here to sell your, your open source software. You contributed it.

You're, you know, 'cause there are other, for instance, I, I was interviewing the, the folks at Broadcom, VMware earlier today, you know, they're the third largest contributor to code to Kubernetes. And a lot of people don't think that, 'cause they think Kubernetes hypervisor, they're separate, but no, they're the third largest contributor to code there now. And I commend them for that.

But let's face it, they're feathering their own nest. Right. You know what I mean?

And, and one can say that even if they're not doing it for that being the only reason, one can still say, well, but they're, they're selling software. Right? Right.

And two it different thing, you are doing it. 'cause it's, it's, it's making what you guys do better, but you want everyone else to be able to have access to that as well, because you're not selling. And to be able Contribute To it, it, and well, that helps Make it better.

Helps. Yeah. Well, it's like paying it forward, right?

Mm-hmm. When you contributed to the community, hopefully others follow. And, and it's all that karma wheel that goes round in m Right.

And paying it forward also has its dividends. Like for Argo project, uh, red Hat is now contributing as well. And they've created cool features.

There's an MCP server introduced, and now we are also leveraging from the community. So like, if we were to do all this ourselves, it's all our resources. Uh, versus now we are getting back from the community.

So the dividends come in, it just takes some time. Absolutely. Yeah, it does.

And that's karma, right? It goes round and round, but it, it, it does pay off. Yeah.

If you don't mind though, I, I want shift, I'm shifting. Um, no, no, but ai, yeah. Right.

Yeah. Everyone, well, not just this time for the last two, three years, you know, we all lead with ai and I, I think just as Kubernetes itself is maturing a little bit, we're, I'm not saying AI's mature, we still have a long way to go. But we are starting to see definitive patterns in how we use ai, where we use ai, who we is using ai.

Let's talk a little bit about Intuit ai. Absolutely. Jamil, yeah.

The last stories you can tell. So yeah, one of the things like our, you know, mission. In our mission statement, we have three statements.

We have, uh, less work or no work, more money and high confidence. Like we have these three statements. And as we think about ai, like it helps you make more money, it helps you do less work, and it gives you more confidence.

And it's very critical to our consumer products that they have AI in them. And that lets our consumers, uh, get all of these things. Now, when it translates to, like in engineering organizations, there's like, you know, we think about it in three areas since we are all like in the platform engineering space, there's one is like, how can AI make platform engineers more productive and make infrastructure awesome.

The second area is how can it help general application developers be faster? So how do developers go faster with all the coding tools and the tools they have? And the third one, uh, is more around like, how, how are we using ai, uh, where we are giving developers ways to build AI applications for customers because they're new tools for building ai.

And, you know, our developers need to know how to use these tools and how to get faster from an app that did not have AI to now having AI in that app. So for all these three areas, we are investing heavily. And maybe I can give you some examples for each of the Area.

I was gonna ask you too. Yeah. So go ahead.

So For the platform engineering side, uh, this is where like, Kubernetes is hard. That's something, that's why platform engineers are paid so much. Uh, now with AI coming into the space, they're actually gonna get more and more productive.

Uh, for Argo CD for example, we added Argo CD assist. And that is when, you know, people used to ping, uh, our platform engineers and say like, what is wrong with my deployment? Now, AI can just build a summary first, and if the summary doesn't work for them, they can still ping them.

Uh, there is build failure analysis where every time a bill fails for a developer who doesn't know what's going on behind the scenes, they get a summary of like, what is failing. And sometimes, uh, there's also remediation, uh, where it tries to remediate the fact. And developers don't even know what is happening.

We call that done for you experiences where things are done for you and you don't even have to worry about things. Uh, so that's on the, like how AI is helping platform engineers. Let, let me stop you right there on the, this platform engineering thing.

So, you know, we, we started a new site since last year when I interviewed you. Hopefully it's on there. com, on the edge, right?

com. Because platform engineering, I think has emerged as its own discipline. Mm-hmm.

Yes. It's closely tied into DevOps. Yes, it's closely tied into, uh, cloud native, but it's its own thing.

Um, when we look at platform engineering, though, I think two years ago, three years ago, if I asked someone who claimed to be a platform engineer, what are you doing? Kubernetes is hard. And we're gonna make Kubernetes easy.

Not easy. I don't think it ever gets easy or manageable. Yeah.

But I think if you ask most platform engineers today what the mission is, they'll tell you IDP internal development platform. Right. Because that seems to be where the action is in platform engineering and also a lot of where the AI is going.

I'm wondering what you're seeing as a real life platform guided Intuit. Yeah. Uh, so Intuit got lucky.

We invested in an IDP way early before Backstage came out. Every third person I talked to at CubeCon, they're like backstage, backstage, backstage. Uh, because the IDPs serve as a, as a web layer to all their abstractions.

So making Kubernetes easy is through abstractions. And where do you put them? You can't put them on Kubernetes itself.

'cause then you have learned that. So instead they have an abstraction on top of that, which is IDPs. Most IDPs are going to have an AI assistant moving forward.

And that's going to be your platform. Engineering team's first line of defense. And if that AI is not able to answer or do things for your, uh, general developer, that's when you go to the platform engineers.

Versus today we live in a world where like the first choice might be platform engineers. So then I think, you know, IDP and AI together, like in one platform is going to be our first line of platform engineering defense. And you'll see more and more vendors come out.

You'll see the Kubernetes ecosystem getting well integrated into it. Right now, it has all the popular projects, but you'll see a lot more projects going into it. Because once you have an IDP, you have to integrate Jira, you have to integrate GitHub, you wanna integrate Argo, and the list never ends.

No, no. So it's a integration. But I, I think we learned something from like APIs, right?

Which is, you don't wanna do a one-off for every single one of these projects or every single one of these products that you wanna plug into your IDP. You know, I, I think with Agentic ai, we have the ability with let's say an MCP server to, to not repeat the mistakes of and sins of the past. Right?

We, we could do that very quickly. Is that something Intuit is looking at? Yeah.

And that's where like if you look, I would say five years ahead, uh, I would also say like, you know, the I-D-E-I-D-P experience might also shift left. We are thinking about shifting left. And a lot of these experiences are happening within the IDE itself.

So when you're developing, you also have a chat agent, a chat window with CPS enabled. So why wait until deploy? And then asking the questions like, you are coding, you're asking the questions in the side, like, why will this bill pass and your MCP, like your IDE, then call the AI MCP, try to run it in your cluster or tell you information about your clusters.

And before you even push the code out, you will know like what's, you know, what's wrong with your code. So, you know, there's a shift left happening where you want to alert the developers before they push it to a stage where they need support. Like things need to be fixed right.

When they're working on it. And I'm hoping that our tool chains also support that. So you have that chat window that provides you alerts and errors early on, then you deploy, then your IDP has a chat window that you have alerts and whatnot.

And then you have all your operations dashboards and those things. Those also have those. So like you will have a, and you know, a lot of the companies are, their jobs are to unify these assistants.

'cause there's like almost too many these days and they don't talk to each other. So you have to ask the same question here, here, and here, and choose the best answer based on which one's the most mature. So, you know, that's kind of what I've seen happening as well.

I get it, I get it. I was gonna make some joke about like, okay, everybody stop building Dev Dov portals and just, you know, focus on IDPs. But I bet you next year when we're having this conversation, that will probably be a reality or much closer too.

We'll see. Yeah. You know what, this, this, you've said it now it's out in the Universe.

Exactly. Don't even Get money. So we'll, we'll come back and, and get it.

Um, so that's AI and developer platforms. Mm-hmm. But there were two other AI use cases for Intuit here.

Yeah. So the other one, uh, which, you know, in the last three to four years, we have seen a 12 or five years actually we have seen a 12 x increase in developer velocity. Uh, and the reason being all of the, like the AI coding tools, there's a boom of that.

Uh, right now the market's changing so rapidly. Every three months there is a vendor. There is cursor, augment code Gemini Codex, you know, more, more to come.

Uh, and, uh, what we adopted early on as a strategy is not to choose one option, but to give developers flexibility. But in addition to the tool, we are providing Intuit, uh, code as context to all of these capabilities. So when developers ask like, how do I write a web plugin?

It doesn't tell you just a how to write a generic web plugin. It will tell you how to write an Intuit web plugin. And that has been like very crucial to our journey where when our developers ask for help, it is, it has all theisms baked into it, uh, to begin with.

Um, so from a, you know, code gen, code generation perspective, like I think I'd advise everyone on the same journey to like not focus on the tools. 'cause the tool market is going to change. Acquisitions are going to happen.

Chad, GPT and the other vendors will release our OpenAI will release new, new tools. And your developers are always gonna say, this is the coolest or that is the coolest. And you can't in a real world support all of them.

So I, I'd see this market will evolve and ultimately you'll see clear winners. Uh, but right now you can focus on making your AI such that it doesn't give generic answers, but it gives like your company specific answers as much as possible. Excellent.

Yeah. Let's talk about something really important. Customers.

Mm-hmm. How's AI helping you with customers? So, uh, in the customer landscape, like there is, you know, there's immense opportunities.

So on the QuickBooks land, like we have certain agents that help people get, get paid five days faster. And that's like very much aligned to our, our mission of more money. Uh, who, who doesn't wanna get paid faster like I do.

Like, you know, so there are things that would reconcile your transactions on QuickBooks faster. And that enables business owners to also like run their accounting faster. Um, so that could mean vendors are getting paid faster, employees are getting paid faster.

Um, so that's on the business side. On the TurboTax side, like if you file taxes with TurboTax last year, it will do gen AI summary and it will kind of make you feel more confident that, uh, because TurboTax is a tool and you file your taxes yourself, you put the documentation. So sometimes you are doubting yourself, like, did I put everything in the right way?

But now AI is gonna say, Hey, you, you put this document but you didn't support it with this other document, that's usually the case. So maybe you're missing to upload your gains on the stock sale or whatever. Yeah.

And then that's gonna make you more confident in terms of being able to confidently say, I file my taxes. Right. Or I did my accounting.

Right. Or with MailChimp, it's like email marketing who, you know, everyone wants help from AI to write emails is in there. Now it also tells you what is the best time that this person's likely gonna open this email.

And that is intelligence. Where before it was like, send an email at 9:00 AM that was the the norm. Right now it's like, send an email to the people when they want it and when they're most likely to open and read this email.

And that is a game changer in Sure is. And, and, and, and, and that kind of capability in the hands of a small business owner. Like if you're a big company, yes you have those capabilities, but a small business owner being able to personalize up to that extent is magical.

Absolutely. To stay on top of all the regulations and all the rules. And you know, imagine if somebody suggested, Hey Alan, like you're leaving money on the table 'cause you didn't write off all these things, but you can write those off.

That's legal. Yeah. And you'd be like, Ooh, let's do that.

Then more money for you. Oh, I love paying more taxes. Especially unnecessarily.

Uh, yeah. Uh, excellent stuff. So, you know, it's funny, I was talking to people earlier last year, we were talking a lot about AI this year.

I haven't heard as much though. It, it's kind of built in now. And um, but we're still just scratching the surface, right?

Mm-hmm. There's so much more it's gonna happen here. It'll be interesting to see how that plays out.

And I'm also interested to see how it plays out from an open source perspective, how much of this winds up on that side by the project pavilion mm-hmm. Right. Versus inside of commercial products only.

So it'll be interesting. It'll be interesting. Yeah.

And the c ncf F is, is working hard. We, as you know, I'm a CNF ambassador and have been for many years, um, over 10 now. I guess I just saw my picture on a slide saying something about 10 years of contributions.

Uh, so Wow. But that's, we do talk to them a lot and we just had a meeting with them a couple weeks ago, you know, really asking that question, like, what's your plan with ai? And you know, we had a bunch of projects that we think would be great to be in the C ncf F and then they told us about some stuff we didn't even know about.

So the conversations are happening. They are talking to the end users. Um, and, and they're also trying to get, you know, the ois get publishing more, more and more standards all the time, which is really good and really helpful.

Agen is a very big conversation in Kubernetes right now. Uh, so that's a hot topic. Yes.

Um, so we're excited, we're excited to see where it's going. Yeah. It's this morning they announced the AI Big Bricks project.

Uh, they're working on Cajun the Envoy, uh, like AI gateway with Tetra and Solo. Uh, those are all like, exciting projects. Uh, and you know, I don't think, uh, like, you know, AI and Kubernetes are like so separate today 'cause people are already running a lot of AI workloads on Kubernetes.

Uh, there's a lot of libraries out there now. It's just like, you know, it's the same way where the coding agents are like, which ones are gonna hear like, we gonna be here to stay and which ones are gonna be just like noise in the space. So we should find out.

And you know, this is a great place to know and feel from all the audiences here. Agreed. Yeah.

Agreed. Well, listen, we're, we're probably overtime to tell you the truth. I want to thank you both for coming here on Tech Junk tv.

You have an invitation in next year, but let's not wait till next year Exactly. To catch up. Thank you.

You always love being here. Okay. Alright.

Thank you for having us. Yeah, right. Thank you so much.

Intuit. Open source Dynamos here at Q Con. We're gonna take a break.

We'll be back in a minute.