In the season four premiere, Amanda Kelly, Product Director at Snowflake and Co-Founder and COO of Streamlit, shares her expert advice on data applications, creating a data-driven company, pushing your data forward, and so much more.
In the season four premiere, Amanda Kelly, Product Director at Snowflake and Co-Founder and COO of Streamlit, shares her expert advice on data applications, creating a data-driven company, pushing your data forward, and so much more.
--------
How you approach data will define what’s possible for your organization. Data engineers, data scientists, application developers, and a host of other data professionals who depend on the Snowflake Data Cloud continue to thrive thanks to a decade of technology breakthroughs. But that journey is only the beginning.
Attend Snowflake Summit 2023 in Las Vegas June 26-29 to learn how to access, build, and monetize data, tools, models, and applications in ways that were previously unimaginable. Enable seamless alignment and collaboration across these crucial functions in the Data Cloud to transform nearly every aspect of your organization.
Learn more and register at www.snowflake.com/summit
Producer: [00:00:00] Hello and welcome to season four of the Data Cloud Podcast. Today's episode features an interview with Amanda Kelly, product director at Snowflake, and co-founder and COO of streamlet. Streamlet was acquired by, Snowflake in the spring of 2022, and Amanda updates us all on the progress gained since then.
In this episode, Amanda talks about the incredible benefits of data applications, how to push data forward, creating a data driven company, and so much more. So please enjoy this interview between Amanda Kelly and your host, Steve Ham. How you approach data will define what's possible for your organization.
Data engineers, data scientists, application developers, and a host of other. Nationals who depend on the Snowflake data cloud continue to thrive, thanks to a decade of technology breakthroughs. But that journey is only the beginning. Attend Snowflake Summit 2023 in Las Vegas, June 26th to 29th to learn how [00:01:00] to access, build and monetize data, tools, models, and applications in ways that were.
Previously unimaginable, enable seamless alignment and collaboration across these crucial functions in the data cloud to transform nearly every aspect of your organization. Learn more and register at www.snowflake.com/summit.
Steve Hamm: So Amanda, welcome to the podcast. Thank you.
Amanda Kelly: Happy to be here. There's
Steve Hamm: a lot of talk about data applications these days, but I have a feeling that not all of our podcast listeners fully understand the technology and also the impact it's having on business.
Can you break it down for us and share why having interactive data apps is so important in today's organizations? .
Amanda Kelly: Yeah. To me it's all about what are you trying to do with the data, right? And ultimately, you know, when when we embark on data, I mean the data science team does something usually that's to inform some type of a business problem.[00:02:00]
right? So sales needs to understand, you know, what's going on in, in X territory product needs to understand, you know, uh, something about the usage of the product, right? It's, it's so that we can inform a lot of our business stakeholders how to do something. And the problem that I see a lot of times is just that there's so much knowledge that doesn't exist in our databases, right?
The knowledge. In the heads of the people that, that are making this decision. And so I have often gotten in an app to me where the data science team will be like, well, we looked at this and, and users are doing this. And I say, wrong. , something's wrong with the data. Like I, I know that's not happening.
Right. And then we dig in together and we're able to find out, you know, what's going on. Or I have given them a request of, Hey, I need to know this number. And then when I see it, you know, I'm like, that's not actually what I needed to know. Right. There's, there's a lot of iteration that goes on, and that iteration goes on because we have a lot of.
Great people doing great work in different things, and the marketing team is, is are not necessarily data experts and they don't understand sales. Right? We, we all do these separate things in separate roles and what a [00:03:00] data application can do, especially an interactive one, is it's a way to bring people together, right around data.
It's a better way to make data-driven decisions. We're enabling the data teams. To surface up the data, the insights that can be used by other people, and then we're enabling a conversation to happen. Oftentimes, I see someone has gotten into an application, we've laid out data, and when they start, you know, sliding sliders and diving down, now all of a sudden, right, the salesperson has an insight that the data team never would've come up with.
because they know something very deep about their territory or, or what happened last week in the media or things like that. And now we're able to have a much better conversation. We can, we can make decisions faster and also we can inform, you know, what we need to have in the data maybe going forward in the future so that we can scale some of these insights.
Steve Hamm: Yeah. So it sounds like it's really something of a collaboration between the data scientists, data engineers on one hand, and the business users on the other, kind of using the application interface as. The place where the collaboration goes on is, is that what you're saying?
Amanda Kelly: Yeah, I [00:04:00] mean, absolutely. You know, in my ideal world, , right.
I wouldn't need any data applications. There would just be, you know, some wizard who would come and they would tell me at all times, what is the right decision right. To make on my product. Yeah. But that's, that's not exactly how it works. And so having a data application, yeah, to me is, is the next best thing.
It allows me to dive in, right? As, as a product manager, as a business user, really understand, you know, what's going on. Formulate a better conversation that I can have with the data team so that we can get to that right decision faster.
Steve Hamm: Yeah. Now just to get even more basic, Yeah. You know, people are familiar with Tableau, Looker, some of these other things, some of these dashboards, and it seems like that's kind of where data apps came from.
Mm-hmm. , but that's not where they're going now, is it? It they're, they're really evolving into something else. So if you could talk about that, that would be helpful. .
Amanda Kelly: Sure. Well, you know, first off, this whole thing is a spectrum, right? Looker and tableau are, are great tools, and I think what they do really great is a lot of scale, right?
So we need everybody in the sales [00:05:00] org to know exactly what, you know, our, our ARR was last week or these things. They, they work very well for well understood problems, but right. The speed at where we have to make decisions right, is a lot of business leaders is so, Right. Oftentimes we don't have the right data or we need to look at it in a different way.
Right? It's not something that's easy to do out of the box. My business is unique, my situation is unique, right? In terms of what we did, and that's where really having something that's a lot more customizable, a data app that is tuned directly to the needs of my business, the needs to me specifically in terms of what I'm doing, giving me more knobs and stuff to actually dive in.
To understand that data, you know, produces better decisions in, in, in a world where, you know, what you value really is the speed this newness, these things, these problems we haven't seen before. That's where I think it's really, really powerful to be able to give your data teams the ability to just quickly make apps that involve other people so you can get to that new
Steve Hamm: decision.
Gotcha, gotcha. And Stream Lit is basically a, a, a platform for [00:06:00] developing data applications. Tell us a little bit, you know, when and how did the company get formed and what's, what's the, what's the big offer that it has for application developers? What does it do better than others? That, that, that it's, yeah, that has really brought it along.
Amanda Kelly: Yeah. So, you know, to, to understand stream it. I'll, I'll take you back 10 years where my two co-founders and I, Tiago and Adrian, we all met, we got put on a project at Google X and at the time we were trying to create a, a natural language processing, you know, ML fueled tool that also had humans in the loop that you could magically kind of like that wizard , you could magically talk to it and say something like, I need a restaurant reservation, you know, this week, you know, for six people.
Mm-hmm. it, it was very cutting edge in terms of what we were trying to do, and we found. Constantly kept running up. Against issues, right? We needed tools. We needed a way to annotate data in some new way. We needed to look at data in something, something that wasn't, that people hadn't done before, right?
Because we were developing new products where our product was, was slightly different than things were on the [00:07:00] market. And we realized that we really needed a new way. Right, to be able to kind of interact with data and, and really to empower the people that were working on the models, working on the data to really self-serve, right, in terms of creating tools and ultimately creating applications for other people to use, really to speed up the development process.
And so, uh, the genesis of Streamlet was, You know, for your people working with data, can we enable them to run a much faster, a much more agile, rapid iteration cycle where you know what you're doing is you're interacting with your stakeholders, whether that's, you know, a product executive or someone from marketing or an external customer.
They're telling you a deep need that they have, right? That's related to data or that's related to machine learning. How can we involve this in a much faster loop, right? So this is not kind of a big, you give me requirements six months later, right? I give you a tool. How can we enable these teams to self-serve and do that?
And so what we built with Streamli and Streamli is an open source library. It's Python based. I mean, what it does is that it just allows people to very quickly, basically, Front end code, right? You can just [00:08:00] say ST button, right? And then a button appears. And what that makes it really, really easy to do is to, to take the things that these data teams are really, really good at that are really good at diving into data, to understanding it, to structuring it, and then to surface that in a UI that maybe they use themselves right as a tool to, cuz some things are just better done visually in a tool.
or maybe they, you know, they ship that to somebody else so they can do this fast at a relation loop. And that's, you know, I, I always say if we have three values, it's streamlined, it's fast, fast and fast. And that's, that's what we really focus on is, is can we get you to that speed to insight faster? If you had that insight today, what decision would you make differently, right, than if you had that, you know, three months from now?
Yeah.
Steve Hamm: Yeah. Now I understand that the interactive part, and correct me if I'm wrong, really suggests that the users can do a little bit of modification or customization on their end without having to take it back to the data team. So talk about that. What, what kind of freedom or power does it give the
Amanda Kelly: users?
Yeah. Well, it, it gives them as much power as you want to give them, is [00:09:00] honestly the answer. What Streamlet really is doing is it's not giving you like three templates, right, that you can plug into. It's giving you a, a full language that you can use to do everything that you could do in Python, right? So you can do everything from setting up apps to, to monitor your data pipelines.
Right. All the way to, you know, I've made a annotation app that I can send to a offshore group of operators to, to go and tag things in the data and we're surfacing just basically all of. Out of the box primitives, the, the buttons, the charts, right, the image displays that allow you to put together whatever it is you need for that specific business application.
So things I see people doing, you know, I see people writing stuff, you know, you can do a lot of right back to a database, right? You can, you can annotate things, right? You can look at data, you can change data, you can be printing things out, downloading really kind of whatever you need to do and whatever you want to enable those end users to do so they can do their job.
Steve Hamm: Alright. So you're, you're kind of giving them a bunch of tools that they can use [00:10:00] and you're kind of, you're, you're, you're making things either simple or complicated for them, depending on their capacity to, to handle more Yeah. More variables. Okay. I, I get
Amanda Kelly: that. That sounds cool. Yeah. I, I always say the, the best and worst thing about Streamlet right, is it's as wide as Python.
So you can, you can pretty much do anything you want. It just depends on how much code you wanna write. Mm-hmm. .
Steve Hamm: Yeah, I understand that in February, streamli reached a different milestone, more than 1 million monthly users on Streamlet apps. So how has the Streamli community grown over the years and why is this milestone significant?
Amanda Kelly: Yeah, I mean it's significant cause it's a really big number and it's exciting to have, you know, really big. But, you know, when we first started Stream Lit, our goal was really about empowering developers. So we were focused a lot on, on Python developers, and we, we still foc, I mean, that's still our primary focus, which is can we make a magical development experience that people like that, that helps them build the things that they need.
App development in general, you know, really [00:11:00] is kind of, you know, it's a two-sided process. It's about both what does the developer make, but also who uses it and how do they use it and, and ultimately the impact. of the work that people are having or, or is whether or not someone is consuming it. And so, right.
When, when we look at viewers, that's really exciting to us, right? It's, it's exciting to see this grow, you know, even faster, you know, necessarily than the number of developers because it means the impact that developers have is growing. And I think that's the thing that a lot of data teams crave for, right?
They're, they're often not, Seen as the heroes, right, in their org. And, and we really think those people are heroes. They, they can and they should be having more impact. And so it's really, really exciting to see the community building things that, that's used, you know, both internally, right, by lots of users and the, the marketing team, the sales, the product teams to have that impact, but also in the world.
And, you know, one thing to always keep in mind with streamlined is, you know, we, we are an open source project, right? And so a lot of what we do is focused on the community. It's focused on, on Python in general and just about what we can do to really, you know, help push [00:12:00] Python forward. How can we help push data forward?
How can we help make it so that if there's a Covid 20 right bug that's coming out, we can help people spread that information, right? And really have impact with. .
Steve Hamm: Right, right. So we know that there are more than a million monthly views on the apps. How many apps are there?
Amanda Kelly: Oh, I think the interesting thing about Streamlets is that they're different.
From other types of apps. So when we say apps, that often means to you, okay, I hired a front end development team and we put together a lot of wire frames and it took six months and we shipped something, and we expect this thing to live for years. Right? With Streamlets, it really is about can we get you something fast, right?
Can we unblock you today to make a better product or sales or marketing decision? And so when we look at. You know, 90% of them often never live past a week. And that's great. Oh really? That's amazing. Because what we like to see is that [00:13:00] people are rapidly iterating. They found that answer and then they moved on.
Right. That is the value. And so we do have a lot of tools that they get spun up and then, you know, they grow and they serve, you know, tens of thousands of people. And there's a lot of great examples of doing that. But to me, the really, really exciting thing is when I see teams who are just constantly building something.
They had a new idea. They built something new, they got value, they answered a question. That tool might have only needed to be used for a week. We had, um, I think, you know, like during the acquisition that we were doing with Snowflake, we had a point where we had to like de-dupe a bunch of things and it was so complicated and we built a really quick tool to, to do this, you know, in a streamlined app.
And it saved us like a week's work of work. We never needed again it after that week, but it had a huge amount of impact in that week. That's what I love to see with apps.
Steve Hamm: That's just an amazing thought to think about how agile businesses can be about decision making, about collaboration. Really, really cool.
Now, I know you have a lot of apps and a lot of organizations using them, so. Let's, let's kind [00:14:00] of drill down into a couple. I, I've heard that you have some really fascinating ones that just are gonna blow people's minds. So give us a couple of examples of interactive data apps that have built by the community.
They're using your technology and they're really making a difference. .
Amanda Kelly: Sure. A lot of our, our funner and cooler examples are, are things that, you know, are public. Right. Um, and that our community has put out there. I'll, I'll talk about a couple of those and then I could talk about a few things that we've seen being built, you know, inside of businesses as well.
One of the things that's, that's super hot right now is chat, g p t. And so, you know, one of the apps that came out on Streamlet earlier this year was an app called G P T. . And the problem that the kind of problem that a lot of companies or academics were having at the time was if people, if our students have access to this, how do we know that they actually did the work?
Right. And they wrote that, that paper, that exam, as opposed to, you know, they just use chat g b T to do it. And so a student outta Princeton, Wrote over a period of weeks, basically an algorithm to do this built in and streamlet, and made it available to the [00:15:00] community and said, test this out. I think this could be a solution to the problems, you know, that we have.
Right? Researchers. And since then we've seen so many other things that are coming out. We have math, G p T, and even like Bible and Torah, G P T, and like all of these things that people are building on and it's, it's really exciting to see this in the community because, What you see, and you, you can see on Streamlet, right?
And in the gallery and just kind of, if you go on Twitter and even search Streamlet apps is just this rapid iteration that's happening as we kind of all are figuring out this new technology together, right? These, these bigger data problems, these bigger kind of things and, and people are taking the work of other people and they're drafting on it and they're building new things.
And that's what I love to see kind of internally too, right? When we see this at companies that have really adopted streamlet is that they're all kind of working together. They're seeing something that you built and saying, well, let me grab the code. Let me build something. I have a different insight. I can build this out in different ways.
And we see, you know, a huge wide range of things that people are doing right. Um, Delta Dental of New Jersey built this, you know, first kind of AA small application to really help their [00:16:00] call center operations. Figure out, just surface more information right to the operators about what was going on kind of in real time with calls.
Show them a little bit of the, the metadata, whether it's the sentiment is trending up or tending down kind of on a call. And now they've built that into a much broader, you know, and bigger thing that, you know, be partly because they're constantly getting feedback from their customer support team.
They're like, can we have this? Can we have this right? And they're just building basically live and as they go, all of the tools that they need that they see in terms of, you know, issues with their day so that they can free them up to do more Interesting. You've
Steve Hamm: given a couple of great examples of really exciting interactive data apps.
Let's delve down into the, in some of the business uses.
Amanda Kelly: One of our really exciting customers is, is JetBlue. And, you know, they're, they're one of, they were an early user of Streamlet. They're an early user of Snowflake and. They're doing a lot of, you know, really exciting stuff to go from the data and the models that they're, they're building right within the Snowflake ecosystem.
And then, you know, using that to do things like figure out the mitigations for flight disruptions, right? Or [00:17:00] provide predictability to their operational planning teams who need to decide, you know, are we gonna delay this flight and what happens there? And so, or do kind of customer personalization so that, you know, their, their customers can have the best, you know, possible experience.
And so I think, you know, That's a really cool example of like, you know, you don't, we don't think about, you know, we, we mostly complain right about what's going on with slides, right? Ah, the thing is delayed in 45 minutes. You don't think about like all of the complexity that's happening in real time where people have to make a decision.
Am I gonna divert, you know, this slide? Am I gonna do these types of things? And so, you know, what an awesome way. To expose to the people who need to make these, you know, it's not the data team who's making the decision, right? About, you know, changing these flights or, you know, providing something new to customers, but to provide that to people to do that.
So that's a really cool example.
Steve Hamm: Hey, um, you, you made quick reference to this, but it's been about one year since Streamli was acquired by Snowflake. We want the update. How has Streamlets technology and how has the strategy evolved since the the merger was announced, [00:18:00] and also kind of what do you bring to Snowflake and what does Snow being part of Snowflake bring to you?
Amanda Kelly: It was a very rapid decision, right, that kind of both companies made and it just happened to be one of those kind of right place, right time decisions where we recognized something in Snowflake, right? That we didn't have. I think Snowflake recognized something, you know, that we had, that they didn't have and, and we saw this opportunity.
To really bring these two puzzle pieces together and what could be a really impactful way for users. And so the problems that we were struggling with at the time, streamli, going from an open source to something right, that enterprises could use, is that, you know, enterprises have a lot of very different needs right than then.
We were used to addressing, they had, they had issues around privacy and security and, and, and governance and, you know, wanting to have everything, you know, in, in, in one place. And we were gonna have to build a huge amount of te. and a lot of integrations really to, to meet the needs of enterprise. And Snowflake had already built all of that.
And, you know, and not just building that, but you know, snowflake has this amazing data platform where we realized [00:19:00] that if we could sit on top of that, we could make our value proposition fast, fast, fast and fast. Right? Um, that we could really help people go directly from that data work that they were doing prior right to wanting to make something in an app to something that could be.
Very quickly shipped and, and, and make this this rapid iteration, this easy cycle much faster. So that was kind of the vision, you know, a year ago when the, the founders of Snowflake and, and the founders of Streamli came together. And it's really exciting that, that we're, we're, we're actually really very close.
Where's some exciting stuff that we're gonna be able to show off at Summit this year in terms of how that integration is going? And what's more exciting to me is, you know, we're past kind of the first hurdles of integration, which means now we're able to open up our. To not just like, okay, let's glue these products together, but what does this really mean, right?
To have this fast application, you know, enabling people to build that on top of their data. What are all of the, the, the much cooler things right, that we can now build together, thinking about this as a joint product. So I'm, I'm really excited for what's gonna be coming up in the next year. . Yeah.
Steve Hamm: [00:20:00] So Streamlet is open source.
So do you have some kind of hybrid technology strategy and business strategy on this, or kind of, can you explain that?
Amanda Kelly: No, absolutely. It's, it's actually quite simple. The open source technology and Snowflake is very, very committed to Streamlet growing in open source. And I would say, you know, even the, uh, the Snowflake founders are often like, let's make, you know, let's make this open source and let's make this open source.
And I'm like, okay, great ideas. Let's, well, we'll figure out how to do it. We now have, you know, we have very big goals, right? To, to make streamline even more useful to the Python and the data communities in terms of putting even more into the open source. And, and we wanna put basically everything into the open source package that you need to build whatever type of apps that you want.
What we're bringing on the Snowflake side is once you've built that app, you have to share that with someone. You wanna get that impact, right? By sharing it with, with somebody else in your organization. That's also not easy to do. That requires a DevOps and an it, you know, setup. If you're building this yourself, there's a lot that you have to think of, right?
In terms of authorization and security and privacy, right? [00:21:00] We're packaging all of that up and we're making it really, really, For these apps, you know, to run in Snowflake and, and to do kind of everything that you want in terms of this broader kind of corporate need of, of, of sharing and, and, and governing and, you know, making sure that's all in one place.
Yes. All,
Steve Hamm: all the things, all the things that are attended with enterprise publishing and enterprise piece of software. That, that's, uh, that's really cool. Now I know that, you know, in the. There's a tremendous amount of interest in, in data democratization. There's a recognition that this is something that is good for companies, it's good for organization, it's good for society, and I know that snowflake and stream lit both hold this value, dear.
Yeah. So talk about how, you know, the combination of the two organizations is really helping to democratize data for businesses in. .
Amanda Kelly: Yeah. I should say that Snowflake is doing, you know, some amazing things on the marketplace side. Right. Which, which Streamlets gonna help be a part of too, in terms of making ready-made apps available to people.
But they're already [00:22:00] doing such great work in terms of helping people get better access to data, to, to share data, you know, with other companies and, and things like that. And really help to unlock this box in a way, right. That, that. Can work with. And Streamlet is, is a part of that, I think, in terms of how do we, how do we help that data be better understood, right?
How do we help that data percolate right into areas that it otherwise would not go into? So, you know, like we talked about in the beginning, to me, if you want your company to really be data driven, then that means every. In the company has to be actively working, right with data, actively working with your data teams.
And so where Streamlet fits into this really is about democratizing that access. In a way. And so I think the magic of what Snowflake and streamline are doing together is doing that in a way that also protects a lot of, you know, what we need to do with security and privacy, right? How do we get those insights out to the marketing person without, you know, giving them full access to the database?
You know, having [00:23:00] to look deeply into, you know, what might be very, you know, secure private data. .
Steve Hamm: Yeah, that's really interesting. Cause you can have all the data in the world and if people don't have the right kind of access to it, the right tools to get into it, it's useless to them. So I can see how the, the interactive
Amanda Kelly: data don't as a, as a product leader, I don't want data, I want answers.
Right. I have, I have problems, I have decisions that I need to make. Right. But to me it's really. , getting that data into a place that somebody can make a decision with, and I think Streamlet is, it's not the only option, but it's a really, really great option. Especially when you wanna move fast.
Steve Hamm: Yeah. Yeah.
Now, you and your two co-founders came into Snowflake. You have different roles. How would you define
Amanda Kelly: your role? . Well, my role is I am a product director of analytical product experiences, so that includes the streamlined open source product, our community products, also our streamlined snowflake integration, but [00:24:00] also kind of a broader portfolio of authoring and editing experiences.
So things that we're doing in in worksheets and things that we wanna do in the future. Generally. I like to think about it. You know, I spend a lot of time thinking about data scientists and data analysts and like what it means to do their job well and how I can help them. Have more impact in their broader organization.
And then we're trying to just, you know, create all of the experience with whether they're a super advanced Python developer and they love their VS code and they love to develop locally, or you know, they're brand new to the organization and they, they just need to, you know, ship something new out there.
How can we meet them basically where they are so that at the end point, right, they're, they're able to do their job better. They've got all the tools that they need right at their hands, and they're able. You know, the great impact with their work that they, we know that they can have. Yeah. Now
Steve Hamm: it sounds like you spend a lot of time talking to customers, whether they're the, the, the data scientist side or the business user side and bringing the specs what [00:25:00] they need back into your organization.
Is that a lot of what you do and, and, and what are you hearing from people today about what they want?
Amanda Kelly: Yeah, I would, I would only talk to customers if that was my option. I love talking to customers and hearing, you know, regardless. It's just, it's so inspirational and, you know, you'll hear common threads about what people are doing, but also, you know, so many companies are just doing so many cool and unique things and it's, it's really cool to have, you know, a week where I can talk to somebody who's doing.
Something, you know, super advanced and quant in a finance company. And then I can also talk to somebody who's in mining, right. You know, and then I can also talk to somebody who's doing drug discovery, right? And, and I think, you know, that there, there's similar problems in terms of like, none of us have enough resources.
We all need to hire more. We all wanna move faster, right? We all wanna do more with data, right? And, and we focus on how we kind of do those things, but the way that everybody's doing it, you know, is, is different. And that's really cool. And it's really cool to have a product that can span, right? To help you know, all of those types of people and.
You know, we have a goal that, you know, every single one of our product managers, but also our data [00:26:00] scientists and everyone, right? They're constantly meeting, you know, not just with Snowflake customers, but you know, people that aren't customers, people who are Python users, people who are part of the broader streaming community, because it's, the future is already here, right?
It's just not evenly distributed. Right. And that's one of the great things about being a tool, right? That's based on Python, is somebody has already. a version of what we can now, you know, take and make accessible to more people. They, they've written, you know, they had a business need and because they're, they're smart and capable people, they wrote 500 lines of code, right.
Where that could be, you know, five lines of code. And so I love just looking at what people are doing and saying, could I make that easier? Could I make that one button? Could that make that one line of code so that that can now be available to, to everyone in the Snowflake and Streamli community. Yeah.
I see the future.
Steve Hamm: Fascinating. Modern age. We live it. Is this what the
Amanda Kelly: future holds?
Steve Hamm: I feel like we're having a really up to the minute conversation [00:27:00] about technologies, , and data technologies, which is really exciting. But I'm gonna ask you to put on your visionary cap for a minute. Okay. Looking out like five years or more, how do you see data application technologies, transforming business or even.
Amanda Kelly: Well, I never take off my visionary cap. Um, it's, uh, just what, which, uh, which flavor of, of visionary, it's called your head, right? Yeah. Mm-hmm. . Yeah. You know, what's, what's interesting is, you know, in, in some ways not a lot will change, right? In five years, like, you know, certainly, you know, fundamental changes, you know, take a long time, right?
And, and we can accelerate that, right? By building, you know, better products and taking the things that we already see. Like, I see a lot of people doing things like building. You know, really sophisticated ML ops pipelines and DevOps pipelines. Right? And they're, and they're building streamlets into that, right?
So that they automatically, you know, are generated right at certain points. And, and then there's really, really awesome stuff that, that's going on there that I would love to be a part of. I see really exciting stuff that's happening in the ML community and just the speed [00:28:00] of, you know, what's being developed there and how that's being brought back into companies.
And so what I'm really most excited about, I think, in terms of the future is how can. , how can we run faster? How can we learn more? How can we basically empower more people at the grassroots level, the people who are, you know, out doing the field sales, who are, you know, sitting kind of, you know, with our data pipelines, right, to do more and, and, and to ship that right to other parts so that we can do things faster.
And I think it's, it's coming right? The revolution is coming, right? It's, it's already partly here because. So many people are learning this in school. You know, other people are, are coming into this and so I think data fluency, people's ability and comfort of, you know, at least doing a little bit of code coming and then we're working to match, meet them right halfway by making things a lot easier and a lot more accessible as well.
Steve Hamm: Yeah, that's interesting. We, we, we talked a little bit about the large language models, the foundation models, things like that, AI stuff. I. When you look back through the history of technology, kind of, it, it has its kind of pause moments and [00:29:00] stability moments. Then it has it, its rush ahead moments and we're in a rush head moment that's almost, that's making, I, I think it's making people's head spin.
Yes. I mean, things are happening so fast, and especially with, with the large language models, and it seems like I. When you look to the future, I mean, is, is machine learning gonna be in everything? Is, is every data app gonna have machine learning in it or, I mean, maybe I'm behind the times, maybe it already does, but, but explore that a little
Amanda Kelly: bit.
Yes. Short answer. I think, you know, even if you don't know it right, you know a lot of things, you know, the same way that most products that you interact with, right? They have algorithm. Right. Who that are backing them. They're recommending a new product to you. They're, they're deciding which, which search result you see in certain ways.
Those are just, those are evolving. Right. You know, and, and they're, they're becoming more Right. You know, machine learning based Right. Than, than just straight code. And I think that, you know, we're, we're seeing that go in, in, into new and, you know, in, in different [00:30:00] ways. But I think one of the things that's always interesting and exciting about technology is it, it, it, it's very hard to predict exactly what's gonna.
Often win these things together. Right? And there's this great example of, of what happened when ATMs came out, right? Then everyone was like, well this is gonna kill bank teller jobs. And what happens is there's now more bank teller jobs than there were back in the seventies when ATMs came out. And that's because what happened was ATMs actually dropped the cost of opening a bank branch.
And it actually, it took away a lot of kind of the rote work that bank tellers had to do in it. And it just shifted the work in terms of what they're doing. And so, you know, one of the things that I see that that's happening, you know, especially with a lot of machine learning models is it is taking away Right.
Some of the work, but, but often it's, it's not it's work that we're happy to give up. Right. You know, I don't like, you know, the work of having to. Been in, uh, 20 minutes trying to remember what I named a document so I can go and find it. So I can look at that insight. Right. You know, I'm, I'm happy to give that up.
There's a lot of things that we're outsourcing, you know, offshoring or things right now, which is more [00:31:00] rote work. And I think that's, that's a lot of really exciting stuff that's gonna be baked in, which is gonna get us a lot more to this editing creative work that I think a lot of us crave to do, which is more about, you know, I have an.
Right. Can I get some answers? You know, can I get some things? But, but it's just gonna enable us to do more of that. And, uh, so I, I'm really excited to see more of that coming in, in lots of broader areas in the economy.
Steve Hamm: Yeah. Yeah. Hey, I wanna talk about opensource, the opensource community and, and model again for a minute.
When I look at a lot of the, a lot of the innovations in the data world, Are based on open source projects, then have a, kind of, have a commercial application of businesses built around them. All those kinds of things. We've seen many examples, which I think is very exciting for entrepreneurs. I mean, you can, you can have it both ways kind of.
So you've done this, what kind of advice and recommendations do you have for, for other people who are thinking about starting something and using the open source [00:32:00] model, but also. With an eye on commerce and a and, and a and a business down the road. Yeah. How should they think about that? How, how should they proceed?
You know, I
Amanda Kelly: think, you know, one of the interesting things about, you know, open source and development and kind of really opening up, you know, everything, right? Your, your code base, your roadmap, right? Is that you get so much more input. Than you would otherwise. Right. So, you know, you will definitely get contributors, but also just it gets used in new and exciting ways, right.
That you couldn't even have anticipated. And if we had made the choice to close source Streamlet, it probably would've rolled out. We would've, you know, rolled it out to the FinTech industry and then we would've rolled it out to retail. Right. And it, and it never probably would've evolved into what it is.
Because we just didn't have enough people looking at it. We didn't have enough people who cared about it. Right. Who were really kind of investing in doing it. And it is a very different way of, of developing a product that that's much more like you're, you're the banks of a river. Right, right. You know, trying to kind of guide and, and do it as opposed to, you know, you're, you're putting certain building blocks in place [00:33:00] and it's very exciting.
I think it's, you know, when you're trying to do something that is about like, . We're trying to, to, to shape how Python can be useful in the world and to enable data professionals. It's very powerful to go about that as an open source strategy. But you know, it's less obvious in terms of how to monetize, right?
Because you know, a lot of, you know, advice that you will get, right? It's, you know, you, you hold back the features, right, that are most valuable, and then you, you put together a business model around that, right? And you, and you make people pay and you have sales cycle. It's, you know, there's a lot of open source projects that have never made money.
I mean, it's, it's, you know, uh, we're, we're really excited about being part of Snowflake, but it's not like we were, it's not like we were generating a billion dollars in revenue right before coming in. We were still figuring out the business model of what we were doing. And so I think it's, you know, it's really important to make sure that.
What you're trying to do matches to, to an open source model. It's important to have investors who really care and believe, you know, in open source, right. And the different types of business models that you have. I think what's what's most [00:34:00] important is just, you know, designing a product that your developers love.
You know, when, when you're making developer products, to me it's, it's give your developers. Everything right that they need. Give them information, give them time. Right. Be a part of your community because that's, that's how you grow specifically as a language, right? And as an open source project is, is by really, you know, embedding yourself in community, thinking about systems and loops and, and what's frustrating for people.
But you, you do have to kind of do that, you know, with a mind to, you know, especially if you're, if you're trying to make money right. As a company, how you're gonna do that. And we got lucky I think with streamline in the sense that there are these two very clear. Right of like building an app versus deploying.
There were two, you know, distinctly and related things to solve, which allowed us to say, Hey, we can basically charge for this thing that, that really is an enterprise need around, you know, how do you, how do you solve the deployment issue? Kind of, you know, and the sharing issue and the collaboration issue within companies and, and that allowed us, and it's still allowing us to pour a lot of love right into the open source side and, [00:35:00] you know, and grow that with the community.
Yeah.
Steve Hamm: Yeah. , you know, we're about to finish our podcast here. We're on the, we're coming to the end, and I just wanted to talk a little bit about something and we've, we've flicked that a couple times this conversation. It's. . We live in turbulent times. You talked about Covid 19. We've talked about how quickly technologies are changing and understanding of what can be done when them is changing.
We've got, I don't even want to go into detail on, you know, like political turmoil and, and, and warfare. We've got lots of things, lots of. Things going on around the world and we have economic uncertainty. So kind of, this is really a management or leadership question. How do you deal with all this, with your teams?
How do you, how do you keep them kind of moving quickly, agile, but also not freaking out?
Amanda Kelly: One thing that I try to lean into and, and my leadership and manage the smile [00:36:00] is, is follow the. , it's, it's to really empower the, the people who are doing the work, right? I, I like to spend very little time in a high tower saying, you know, let's do all of this and I'll check back it up in six months.
I, I like to see really fast iteration speeds. I like to get, you know, things out quickly, right? I like to empower. If somebody comes to me and says, I have, you know, I'm really passionate about this idea. I like to find time to make that happen because, you know, passion is. Right. It, it tells me something.
Maybe you can't articulate it, but it tells me something about a problem that someone's working on. And it's, it's constantly a balance because, you know, we, we do have long roadmaps and we do wanna ship, right? And we do, you know, wanna make sure that we're on this track. But I think that you also really need to stay open to what you know, your employees are telling you also your, your community, what your users are telling you.
And stay flexible to that. And I think that ultimately that makes for a more. Work environment. When you're constantly learning, when you're seeing ideas, when you're able to participate, write in a process that that sees you, that sees that you see this [00:37:00] problem and you're able to have a conversation that week about, is there something different that we can do?
Right? Because there's almost always something different that you can do. Maybe I can't change the underlying architecture, but I could write an article that helps people understand this. I could put something in documentation, I could put a toast in the product that directs people somewhere. Right? You know?
There's many things that you can often do, I think, to make your users' lives better, and I think that mostly what people are craving, you know, at work is, is they're craving to make a. Right. And you know, you, you might have formed a company that's, you know, figuring out r n a discovery and there's purpose there, but I think that most of us get up in the morning looking to help the person sitting to our left.
Right. And to get a pat on the back from the person on the right. And that's what I love about working on a product that allows people to rapidly iterate together, right. Is I'm, I'm allowing those people to serve each other right. And, and to have ideas and, and to think and see the impact that they're having every day.
And to me that's, you know, that's why I get up in the. . Yeah. Oh,
Steve Hamm: that's, that's a wonderful answer. That's very inspiring. So this has been [00:38:00] fascinating really. And you know, I did interview your co-founder Adriana about a year ago, and I found that fascinating, but I didn't really understand, I didn't really understand where this was going and how powerful it could be.
And, you know, I've been a, a, a technology writer and a business writer for many years, and people talk about the Agile organization, and I really think what you're doing is an absolutely essential. Element, a tool for agile organizations. And I think we, we keep getting, getting better and better tools, but I think this is a big step forward and I think, I think a lot of people want to know about it.
And so it's great that you're out there talking and listening as well. And I think this is gonna be a good thing. Yeah. Well, I'm,
Amanda Kelly: you know, if, if you're listening to this, I encourage you to try it. I always say it's, it's hard to tell, it's hard to tell what Streamlet can do for you because it does something.
I think for everyone else. Right? Because it, it really comes down to what is that thing that if you knew today, if you had an [00:39:00] idea that that really could change, The decision that you're making, what does it feel like to have those answers today? To have that tool in your own hands and so, yeah. You know, oftentimes I feel like I present on streamline.
People are like, maybe, but then I see, you know, one person in the audience, they go and they make something and then that starts a mini revolution, right? Oh my gosh, you made this. Could we make this thing right? And it starts to change. I've seen sales processes change to start with data apps where people are doing live what if, and ROI analysis, right.
You know, to get like a media account. And it's, it's just really exciting to see what happens. When you empower the people who are closest to this to, you know, I think to, to work more rapidly with other people in the organization. Yeah, yeah. No,
Steve Hamm: that's great. So thank you so much for being on. It's been great talking to you.
Yeah,
Amanda Kelly: absolutely.
Producer: Are you interested in learning how to build on Snowflake? Join other developers, data engineers and data architects at Snowflakes Build dot local event series. Roll up your sleeves and explore the possibilities of building on snowflake with local in-person instructor-led workshops taking [00:40:00] place across more than 30 global cities.
Now learn more and register at www.snowflake.com/build.local.