What's New In Data
A podcast by Striim (pronounced 'Stream') that covers the latest trends and news in data, cloud computing, data streaming, and analytics.
What's New In Data
From Data Pipelines to Agentic Applications: Deploying LLM Apps That Actually Work
Spencer Cook, Lead Solutions Architect at Databricks, joins to unpack how enterprises are moving beyond hype and building practical AI systems using vector search, RAG, and real-time data pipelines. He and John Kutay get into what it really takes to serve production LLMs safely, avoid hallucinations, and tie AI back to business outcomes—without losing sight of governance, latency, or customer experience.
Follow Spencer on:
What's New In Data is a data thought leadership series hosted by John Kutay who leads data and products at Striim. What's New In Data hosts industry practitioners to discuss latest trends, common patterns for real world data patterns, and analytics success stories.
Welcome to what's New in Data. I'm your host, john Coutet. In this episode, I'm joined by Spencer Cook, senior Solutions Architect at Databricks. We dive into how enterprises are deploying generative AI and agentic applications at scale, the critical role of real-time, high-quality data, and what's next for streaming and RAG architectures. If you're a data engineer or just curious about building real business value with AI, this one's for you. Let's dive right in.
Speaker 2:Spencer, how are you doing today?
Speaker 3:Spencer, how are you doing today? Doing awesome. I've been looking forward to this, so excited to be able to do this. You know, sleep together right before the holidays.
Speaker 2:Yeah, perfect timing going right into the holidays, wrapping up record-breaking years for both Databricks and Stream, and what's New in Data. So you know just a lot to be thankful for this holiday season, spencer. First just tell the listeners about yourself.
Speaker 3:Yeah, absolutely Thanks, john. So I have been kind of at Databricks coming up on four years now. In June Before that I was pretty deeply involved in the Azure data space generically and at the time there was a lot of web apps, but also data management and obviously Azure Databricks, and so that's been my background and that's a lot of what I do at Databricks. I mostly just help our financial services customers in particular leverage our platform in the cloud to solve their problems.
Speaker 2:Absolutely, and you know this is going to be a fun episode because I love having guests who are always really in the weeds of things and can speak to real world experiences, either building these tools for the product or building these tools in the field for customers, where it's really deployed in the real world. Both of them have their own unique challenges and yeah, yeah, so you've done some awesome work, specifically with LLMs and actually getting business value at large enterprise scale. How are you working with customers to actually innovate with ai and lms in a way that's showing visible business value?
Speaker 3:yeah, so like there's a lot of talk about the types of use cases that we're seeing business value from, I think, with lms, and it's a lot of it's around information retrieval, but then also coding assistance generically, and you know that's all fine, but getting there is really challenging still, and it tends to be the same process that we saw with traditional ML or even doing analytics, where a lot of it is around getting the correct business processes in place, cleaning and organizing your data and basically getting reliable data into the right systems at the right time to be able to do these more sophisticated techniques. It's almost like LOMs are the dessert and you stop like eat your vegetables and your protein and that's the nice reward at the end.
Speaker 2:Yeah, that's a great analogy. The making the data accessible to the LLM and you know that's, you know, sounds easier said than done. Really, everything around large scale data management, indexing the data, make the data, chunking it for LLMs and vector storage and all these challenging things that go on in the background to really make it so the data can bubble up to the user as a chat-driven experience. We all know chat GPT, but a lot of enterprises are trying to do some form of chat with enterprise data in one way or another, either internally for knowledge bases or externally for support or customer experiences. Like what are you seeing in terms of real world adoption there?
Speaker 3:Yeah, I think that there is real adoption.
Speaker 3:Um, we are you know, I personally have been involved with a lot of new states that are in production in various stages throughout this year um, I think that a lot of what companies are trying to do is figure out basically what they can do, take value while still prioritizing safety, and not like putting themselves kind of in a position where they're over their seats and they're like on the front page of the Wall Street Journal kind of a thing.
Speaker 3:We even like in a lot of our presentations internally, when we talk about data intelligence platforms, we show the use case where the guy was talking like a Facebook chat for a Ford dealership and he's like you know, you're going to give me a car for $1, you know, doing basic, prompt engineering, and so I do think that that caused some initial skepticism and maybe even fear. But we, because we have kind of established primitives for things like guardrails and, uh, putting security and governance on top of loms, I do think we're kind of in this second wave of, you know, enterprise adoption and we're kind of doing it in the correct, reliable skill that's excellent.
Speaker 2:what are some of the things that are required? Like you mentioned, you know, ai can hallucinate and give users the wrong answer and make them think they can buy a car for one dollar, as an example, you know, which is definitely in the realm of probably, you know, uh, relevant possible outcomes, especially when you're dealing with these kinds of probabilistic models. So now to make this real, like, how, how do customers deploy LMS and Jenny I in a way where it doesn't hallucinate, especially with with customers, uh, as the end users?
Speaker 3:Yeah. So, uh, a great uh use case that a lot of us have, I think, been accountable for is like where, say, the CEO is your customer or some like internal leader answers where a ai engineer and inside of this genie space basically ask prompts, get answers back, etc. And as it's generating sql code to answer these questions, when you say, hey, that's spot on, that's how you calculate fiscal year over year, you can certify that as an answer. And so when someone gets back a response later on a non-technical user, that shows up as certified and kind of stamped with the seal of approval that came from that person.
Speaker 3:But the LOM is still contextually applying it to questions that come up, applying it to questions that come up. And so if you think about applying that to external users, customers can widely vary Like a CEO tends to be, at least like an authority on, like the domain of the company or whatever else With your customers. That can vary a lot. So basically, you use standard MLOps techniques that have been augmented for LLMs to collect that type of data and basically pursue those kind of correct answers for all the different domains, and so we can do a lot of things to help with that, like help you generate synthetic data, help you track all these different responses that are going back and forth. But that's generally the idea all these different responses that are going back and forth.
Speaker 2:But that's generally the idea.
Speaker 2:Yeah, then this is one of the areas where people are trying to align on best practices for LLM ops, and some of this already exists, with borrowing concepts from ML ops.
Speaker 2:I had two great guests who got into the weeds of this On previous episodes Avi Aryan, who published a book on LLM ops, and then Andy McMahon, who published a book on machine learning operations with Python. And there are so many ways that you can always sort of fine tune and come up with deterministic filtering for LLM results. Some of the vector databases also have capabilities to pre-filter results in a deterministic way. Before doing the LLM-based, whatever it is, nearest neighbor search on the vector format which can give you the directional correct answer on the vector format which can give you the directional correct answer. So enterprises really have to architect their solutions with some of those best practices in mind, especially before rolling it out. So when you're working with customers on this because you have all this extremely innovative technology all under the Databricks umbrella with with ai and data management like, how, so like these lm based applications are, you know they're only as good as the data they're. They're trained and run on.
Speaker 3:So how can companies use their data to to power real customer facing ai operations yeah, I mean, uh, I I think we need, like probably a couple of sets, like maybe we'll do a series on that one of these that's like at a high level.
Speaker 3:I think that by creating whether you call them data products, whether you call it, you know, a gold layer at the end of the day feature store tables, vector store databases they're very similar to facts and dimensions from the old days with BI reporting, and so you want reliable, scalable processes that can basically hydrate whatever that layer is, with reliability and also ideally really low latency, viability and also ideally really low latency, because once you get the data quality problem solved, then all of a sudden people start being worried about freshness.
Speaker 3:So they're like this data is great, can you get me more of it sooner? And so what we see with LLM is kind of interesting in particular is you take those same processes, maybe they're well-established for relational data, like it's easy to write a merge condition, but now all of a sudden, we're managing documents, right, we're managing images and video and all these different things. So one of the aspects that has been interesting about Spark and Delta Lake which a lot of our platform is still based on, that drew me to it early is the abstraction that we can provide over those things when it's like a document. Let's extract that into bytes or let's extract it into tokens and a column, and then we're going to apply the same kind of primitives in Databricks that we would do for other types of AI processing in the past. You're kind of standing on the shoulders of giants not developing a whole brand new process for this new type of AI.
Speaker 2:Absolutely so. What you're really alluding to is that a lot of these fundamentals and data management, data processing and things that a lot of data warehouse architects have applied for years is that is also applicable to high quality, ai on accurate data, because the data can be pre-aggregated, pre-filtered, cleansed all the things that you would do going from like if we were to talk about it in the simplest form. You mentioned facts and dimensions, the you know uh, which is a great way of uh organizing your data warehouse. You can do things like star schema. You can do things like medallion architecture, which is very popular.
Speaker 2:Uh, you know, basically, you have your, your raw tables, which you know those might be your normalized data database tables, your, your raw data coming in from APIs, those documents, things that are just completely not really queryable from an analytical perspective. They're more representing the underlying application they were sourced from. You're pre-aggregating computing, running some compute to filter that data, to join it, to make the columns make sense, add the right metadata and, ultimately, this is something that you can throw in L. You turn it into some data model that you can throw an LLM at and an LLM will be able to make sense of it because the column names will be human readable. There'll be metadata there. All that is really positive for those who have strong data engineering fundamentals, both individually and as a company.
Speaker 3:Oh yeah, I mean, I think, one of the cool things about our conversations over the years. I think we did one of our first webinars a couple months after ChatGPT was released, and so we're talking about, you know, the things that Stream can provide, the things that Databricks can provide, towards things like DI or analytical reporting, ml, and now it's kind of all the same stuff that's pulling us towards LOMs. I think what really drives this home is, with agents, which is this new trend that's trying to replace RAG as the hot architecture. You need to be able to pull answers from not just your documents, but also from your old analytical reporting, like you might need a SQL query to help an agent answer a question, and so all that stuff that was old school now is a part of your LLM system as well.
Speaker 2:Yeah, absolutely, and I came across a really interesting post on LinkedIn actually from Eric Elson, and he's done some incredible work with AI on LinkedIn actually from Eric Elson, and he's done some incredible work with AI and LLMs. And what he said in his LinkedIn post is the returns from increasing compute and model size tend to be logarithmic. That means that the next 1,000x will give us less than the last 1,000x and much less than the last a hundred million X. But all is not lost. Models will continue to get better at being useful to humans. In fact, we've only started down this path.
Speaker 2:We're entering an exciting era where the most interesting problems are at the interface between the models and the real world. And he wraps it up by saying he's excited to work on these problems at Databricks. So it is really all coming together where the inference time is where a lot of the innovation and IP and deployment will be focused on at this point. I think this does tie back to what you're mentioning around making sure the data is properly modeled in a way that LLMs and even small, more precise language models can work with and essentially come up with the right way to summarize or action or generate text from that data.
Speaker 3:Absolutely. I think one of the challenges, like maybe a next kind Excel report to agree with the entire BI report knows how complicated that is. But at Databricks we have a tool called AIBI that essentially is trying to thread those two things together to deliver that kind of an experience.
Speaker 2:Yeah, tell me more about AIBI.
Speaker 3:Yeah, tell me more about AIBI building visuals. What's interesting is, when you click on any of the visuals, it's a text-based prompt so you say this is the data I want back and it decides whether to do a bar chart, line chart, whatever it writes the sql aggregation for you. There's no you know dax or mdx here. And then what's really interesting is, once that's done, you get a button that allows you to just ask questions of the same data model in like a natural language and that's what I was alluding to earlier where those responses you can certify, you can build them into reports of your own the same data model that's backed by you know catalog. You get this ability to ask questions in both a natural language frame or a traditional BI frame and they're both going to return the same answers against the same data.
Speaker 2:Yeah, this is mind-blowing in terms of the amount of innovation here, where now, with applying AI to BI, you're able to automate more insights. And you know I think there's been a few people comment on kind of like point and click bi experiences might get replaced by more low code. Have gen ai go, you know, take, take a natural language query and just run it against my data and then give me the results in a way that you know I can best interpret it, either via charts or just summarizing the data for me in plain English. And it's sort of one of the interesting areas I will continue to see evolve.
Speaker 2:I don't think the standard report is going away, but when people actually look at reports, the follow-up question is always you know, what does this data actually mean, right? So I think having the AI BI experience where it can just summarize the data for you and summarize the insights and and give add perspective to it, that's actually something that lms are are good at, assuming that the underlying data is has all the context that it needs, which is the tricky part, which you know. Of course, data bricks and and stream both work on this problem respectively in different ways. So that goes to my next question there's batch-driven AI models, and then you have continuously updated real-time AI systems that use incremental inference on data that's constantly changing. I'm curious to see what you're seeing out in the field in terms of what's being deployed and best way to work with real-time data versus batch data for AI.
Speaker 3:Yeah, I always love the batch versus real-time question. I hope that we always kind of get to talk about this within the data space. Batch for AI is something that I don't think is ever going to go away, because there are a lot of use cases, almost like in the customer 360 or experience enrichment space, where, like they don't change that frequently, but you need a lot of them. So maybe you have 300 million customers. You want to give them all a custom, you know background when they log in that doesn't need to change hourly, right, maybe it doesn't even need to change weekly. So, like one of the things that databricks that we just launched, um, was a much faster batch inference system behind our AI query tool for Databricks SQL, and so basically what it allows you to do is take any model against a giant batch of data and just do inference in a highly accelerated concurrent way Classic Spark, right, and so we're still excited about that. We're still innovating in that.
Speaker 3:We also were one of the first commercial platforms to offer a continuous find, which basically means you have a model like llama, whatever else. It kind of knows uh, all the all about the internet, all this public vocabulary. Hopefully it doesn't know about your IT and so your subject matter expert terms. Yeah right. So basically in say, like an internal chat, you can be feeding it tokens and vocabulary and project names and sprint names and all this different stuff from your internal system, and it just is picking that up as vocabulary, almost like just how we pick up language in the real world. So I think you'll see both continue to be relevant. But Databricks has some pretty cool innovation happening in both areas too.
Speaker 2:Yeah, absolutely, and the continuous fine fine tuning is certainly an area that data teams are continuing to explore.
Speaker 2:The interesting part here is, with RAG, you can have these larger and larger context windows for the model to work with and, on the other hand, what you can do is fine-tune and update the model directly, so you see cases like that and then you can update the data that's accessible to the model. So you know like, for instance, when we work with UPS, they're working with streams who have some AI-driven policies to protect packages that are delivered, which ultimately comes down to better customer experience because the package is more likely to arrive at their doorstep. They branded the solution as battling porch pirates and battling porch pirates, and that's one of the examples where they're taking the real-time shipment data, claims data you know other types of data pulling it into their delivery defense system, which is powered by AI and that has to rely on the real-time operational data coming in. So that's one of the other things that you know from their perspective. You know they work with Stream to just get it in, make it, pull in the new data and make it accessible to the model, rather than fine tuning the model itself.
Speaker 2:It's kind of it's a form of rag, right? Yeah, We've augmented generation. So lots of interesting approaches. What do you see as being most adopted from your perspective?
Speaker 3:You know, rag still continues to dominate. Um, I I do know of a lot of customers that are starting to adopt agentic systems. Um, we have, uh, some public use cases I think it was at day that I summit where we're talking about query language, where it's basically this you know, know model that does text to SQL, but for their proprietary query language within the Fax tool, and they're basically combining commercial frontier models with more foundation models that are fine-tuned on their tokens, like I alluded to earlier, where it knows their different types of example queries and they're combining that in an agentic way to basically leverage best each type of system and ecosystem. So I think that that will continue to rise To your point, though RAG is kind of.
Speaker 3:It never really went away, but it's having a new emergence in the way that we can add new types of augmented data. So, instead of, like you said, not everything has to go through a vector store, that can actually be kind of bad for latency. With the ability to query directly off of a stream, for instance, you don't necessarily need to know the vector embedding relationship between those terms. They're in the same topic on a stream and there is a natural kind of temporal relationship in the streaming events, and so you can pull something like the most recent customer in the last five minutes directly into a RAG response without it needing to go through a vector database and all these different things. So I think that, to your point, rag and streaming are a really cool match and I think we'll continue to see innovation in that area.
Speaker 2:Yeah, of course, of course. And you know, and we're always excited when we have a joint customer that's using both Stream and Databricks, supporting hundreds of streaming connectors into Delta Live tables, which is a great way to ingest data in a way that tracks incremental changes and bring it into the data lake and then from there, you can run all your processing within Databricks, like we said, prepare that data for analytics and AI use cases, do all the joins, the data modeling, setting up your facts and dimensions, setting up your gold tables, your platinum tables Big quality roles in there, absolutely A lot roles in there, absolutely A lot of powerful stuff. So the other thing I want to ask about is have you seen examples where stale data can impact AI accuracy and decision making?
Speaker 3:Yeah, uh, say that one more time uh, stale data meaning that's oh stale data.
Speaker 3:Sorry, I heard sale. Yeah, stale data. Yeah, no, you're, you're great. Uh, stale data. I I'm sure that there is, uh, you know quote like from mark twain or something about this, but it's like misinformation can be more dangerous than information, right. Where it's like, if anything, if you're going to return a stale result, it would be better to just say I don't know, I need to refresh, right. Oftentimes they're giving the wrong answer, and so I think in a normal data engineering system, normal analytical process, you might even design it that way Like you might have time to live things like that Often in LOMs, where you're presenting just a blunt paragraph of information to the user. That's context-free, right and so I think stale data can be even more dangerous than ever when we think about these types of AI systems.
Speaker 2:Yeah, definitely. And especially if you're looking at customer-facing experiences where someone wants let's just take airlines, for example I made a change in my reservation and I want to go into a chat experience and say, hey, give me my latest boarding pass, right? And if, whatever the chat experience is behind the actual core reservations database, it's going to give them the wrong boarding pass or give them some sort of error, right? So this ultimately you know, stale data ultimately materializes in poor customer experiences, especially when it's more operational and the analytical use cases where you're summarizing your annual sales performance. Sure, of course you want to make the right trade-offs and say, okay, there we can rely on more batch processing and less on incremental data. So it's always about looking at the use case.
Speaker 3:Absolutely incremental data. So it's always about looking at the use case absolutely and like people uh get excited about things like lineage and audit trails and this different stuff. I think a lot of that it's good for uh compliance. Hopefully you're only getting audited, you know, once a year like in a predictable manner, unless things go south. I think day--to-day kind of operationally lineage is really powerful because you can better understand the relationship between tables. But if something is stale you can kind of understand the production line upstream and hopefully optimize it, work with that team to get your data faster and things like that.
Speaker 2:Yeah, absolutely. And the other part of AI bringing it into enterprises is, of course, another foundational element, which is data governance. So how is data governance impacted by AI?
Speaker 3:Yeah, I think you can see it as like a blocker or an accelerator. In my opinion, it can really be an accelerator because, like an engine, when you have lineage, when you have an audit trails how people are using all the different objects in your data platform. In the Databricks case, like this isn't easy, but in Databricks we make it easy. You can basically create what we term data intelligence, where you're combining that metadata as basically another rail of information that goes into an LLM. So when you ask the assistant like, hey, find me a table about X, it knows that you're part of this team, it knows the other tables you've accessed in the task, etc. So we try and bring all those things together.
Speaker 3:But I think it also has made it so that we can innovate in areas that a lot of maybe AI tends to struggle, which is like regulated industry. I was just at Money 2020 a couple months ago I think now and MasterCard did a great press release where they have a Gen AI assistant platform helps out customers with onboarding. But one of the things they highlighted was basically, by deploying it on Databricks, they had more comfort with things like governance and that allowed them to basically innovate faster because they were able to be comfortable with governance, being kind of a first-party citizen in that data intelligence model.
Speaker 2:Yeah, that's a super powerful case study. And, of course, money 2020 is one of the great events in the finance industry. I believe it's in Vegas this year, right?
Speaker 3:It's always in Vegas, like the big Venetian Hall, and it's so interesting. It's like the kind of way to see what the sort of fintech meme of the year almost is, where you can kind of see the trend that everyone's going to be talking about in 2025. That was sort of the hot topic of, you know, money 2020 and 2024.
Speaker 2:Yeah, definitely going to be an area where we're going to see continued innovation and adoption in the enterprise and especially, bringing business value back. That wasn't possible before. Ai just the speed at which you can iterate and launch transformational products with data and AI. And then, yeah, making it easier than ever for data engineers to really have leverage and bring value to the business, because before data engineers were mainly focused on bringing reports online. Now it's these AI driven applications.
Speaker 3:It's so funny you bring that up. I was a data scientist by training but I kind of, you know, got into the world for the first time where you have, you know, dirty data it's not, you know, being prepared by your TAs, you know, and turns out there's way more data engineering problems than there were like ready to be action on data science problems. If you look, say, 10 years ago to be actioned on data science problems. If you look, say, 10 years ago, and so that's kind of what got me into data engineering and a lot of these areas like governance in the first place is to try and create that you know era where we can finally do data science at scale.
Speaker 2:you know, like we've kind of all been dreaming up for a while yeah, and definitely augmentinging all the human work required for data science and I've worked so many data scientists who just end up being data engineers because that's where 99% of the work is actually required to make the data usable, you know so this can only be accelerated by AI.
Speaker 3:This can only be accelerated by AI. Yeah, I think we had a pretty interesting article like a year ago and then we updated it recently. It's basically like what do these assistant tools and kind of what does a Databricks assistant mean for data engineers? It is pretty interesting. I really think that it's not disrupting people's careers. I think that, if anything, this lets you focus on this stuff that you care about as a data engineer instead of the stuff that you would consider like outside your role or like boilerplate. So I think you know my advice to any sort of practitioner working with these tools is like embrace it, use it as a way to accelerate. You know what you're doing already, but they can really be great tools.
Speaker 2:Absolutely, and it's going to be really cool to see, especially how data engineers are able to move fast and adopt these AI-driven workflows. Move fast and adopt these AI-driven workflows it does sound like a lot of the innovation ahead of us is going to be mainly centered on the actual insights and getting the actual value, either through applications or data science, and finding these insights on vast amounts of unstructured and structured data that's hard to make sense of. So, yeah, it's a super exciting time right now. I think we'll definitely, if we do this podcast again in another four months, we'll probably have tons of other exciting stuff to talk about.
Speaker 3:Last time we yeah, last time we talked it was on the TWI webinar, and I think that might have been six months ago, and even since then things have changed so rapidly oh, it's kind of ridiculous if you actually take yourself back, um, you know I I try and not stop and think about it and just enjoy the ride, right, but I know, especially at Databricks here in the next four to six months, you guys wouldn't believe what we are cooking up. So it'll continue to be exciting coming from our side, for sure.
Speaker 2:Yeah, absolutely. And now when we work with joint customers, it used to be all about getting data into the lake and making it accessible for analytics and reports. And there's amazing use cases there as well. And they actually showed the reports. The executive facing reports every airport and the average maintenance time for aircrafts there. It's really their operational dashboard for tech ops and flight operations and that's all powered by you know, streaming the data into the lake house and running those reports on top of the Databricks warehouse and that's powerful.
Speaker 3:It's a total. Yes, sam, like the way I try to save it at people that have budget power is like they're, you know, think about. Like you know, you can't create or destroy math, right? So your IT budget, let's say it's fixed, right? Well, the only way that you're going to come up with money to spend on all these awesome new gpu hours you need to, you know, host all these amazing models, is by freeing up budget kind of elsewhere. And so I think we do that through the stuff you just alluded to continuing to pursue lakehouse architecture, continue to replace, you know, stale out of the etl systems with modern, you know, streaming uh pipelines on streamless data bricks, like. I think there is a lot of potential there. And the point of all of it, it's not just like taking that money and putting it in the safe, it's taking it and putting it right back in the coffers for your LOM projects and using that as an accelerant.
Speaker 2:Yeah, absolutely, because there really is a lot of ROI for these AI-driven initiatives like what you were alluding to. Especially, the time to value for these data engineering and data science projects is significantly accelerated. And then, of course, the art of the possible thinking of what you can actually do with customer-facing experiences to innovate there. I mean, I know it's very early, but you look at, like Apple intelligence and people are saying people have opinions on how that works at this point. And people are saying people have opinions on how that works at this point. But if, when Apple is rethinking their whole UX to be around intelligence and AI and partnering with OpenAI and ChatGPT to make that a native part of their product, that's the future of UX. Right, it's AI-driven experiences. Now, every company that has a mobile app, every company that has a website, you know in every industry. Now you have to think about, well, what's my ai, intelligence driven, uh user experience going to look like? Because that's what everyone's going to rely on in the next few years absolutely.
Speaker 3:I think apple, uh, is kind of validating our approach because in their case, it's it's you know kind of validating our approach because in their case, it's it's you know kind of your local device. It's not a data lake or a VPC, but this idea of taking a foundation models and frontier models, combining them with your private data you know local way and using that to provide enriched intelligence. That's, that's a data intelligence platform. So you know, I'm a I'm super excited to kind of have that in my pocket as well. It will be validated better perhaps.
Speaker 2:Yeah, yeah. So now, similar to how enterprises had to build their iPhone and Android apps and their web experiences in the time of the internet, now they'll all have to build their intelligence-driven user experience, and that's where these AI data platforms like Databricks and then ingesting the data with Stream, it's going to be table stakes in every enterprise architecture. So it's an exciting time.
Speaker 3:I think we are very close to you know, in the marketplace Databricks. As a marketplace, you also have App Sounds similar to App again, right, and so I think we're very close to this reality where you, you know, click a button in the marketplace, it spins up an application that's pointed to something like Stream on the backend and the customer all they see is this awesome AI experience on the back end. It's connected to this rich, fast data. They don't need to care about those details and they're just off to the races.
Speaker 2:Yeah, 100%. It's going to be an exciting future for sure. Spencer, where can people follow along with your work?
Speaker 3:Yeah, absolutely so. I'm most commonly posting on LinkedIn. You should just be able to find me, spencer Cook, databricks, and then I also contribute pretty heavily to the Chicago Databricks user group. So you know anyone in the Chicago and region. Definitely come check out our events and then I'm always at. You know, data and AI Summit, money 2020, the big show. So please, you know, reach out. It would be great to you know. Chat more about these topics.
Speaker 2:Spencer Cook, great having you on this episode of what's Soon Data. We'll have the links out to your linkedin and for people to follow along with your story going forward, and thank you to the audience for tuning in yeah, thank you everyone.
Speaker 3:Thanks john.