#347 Democratizing Data Access: Blake Burch on Low-Code Solutions and AI

Mehmet: [00:00:00] Hello and welcome back to a new episode of the CTO Show with Mehmet. Today I'm very pleased joining me, Blake Burch. Blake, the way I love to do it is I keep it to my guests to introduce themselves to us. So tell us a bit about you, your background and what you are up to currently.

Blake: Yeah, thanks for having me on the show, Mehmet.

Blake: Um, I'm Blake Burch, I'm the co founder and CEO of Shipyard. We are a modern data operations platform that helps teams be able to move their data end to end, either in a low code fashion or by writing your own code. Um, and I kind of came up with the product, uh, through my experience working as a, Uh, the head of data over at PMG, it's a digital agency where I got to work with brands like OpenTable and Sephora and Gap and Cirque du Soleil, um, all sorts of different brands figuring out like how they moved and use data in their organization.

Blake: I just found that it was really difficult. So trying to make it easier, uh, for folks along the way.

Mehmet: That's cool. And again, thank you for being here today, Blake. Now, [00:01:00] the first thing I want to ask you is like, before you start GPR, like you started, you created actually. More focusing on marketing and then you shifted to technology, teaching yourself SQL and Python.

Mehmet: What attracted you, you know, to this domain and what kind of challenges you faced during this transition? And of course, how did you overcome them?

Blake: I don't think there was necessarily something that attracted me. Uh, initially I think it was like a necessary evil Because when I was doing the marketing work, uh, I I kind of found myself constantly pulling things into excel Um, I was always downloading the same data.

Blake: I was always filtering it in the same way Uh, and then I was always taking whatever I had created, making like a bulk upload sheet, and then uploading it to some service so I could accomplish some very, uh, specific action. Um, and to be honest, in the early days, um, I got too good at Excel [00:02:00] formulas. Um, like, I'm almost ashamed of how much time I spent with some of that stuff, to where I'd have, like, 12 to 15.

Blake: 15 line, uh, Excel statements. I'm like, okay, this, this is unwieldy. There's gotta be a better way to do this. Um, so ultimately I kind of, uh, taught myself SQL and Python just from trying to figure out how can I stop using Excel and doing this manually? Okay. How do I download the data and put it into a warehouse?

Blake: Okay. It's in the warehouse. How do I use SQL? SQL to grab the data and automatically run these filters on a daily basis. Okay, great. The data looks correct. Now, uh, how do I like send that back to a platform? Oh, I have to use Python. Oh, I have to learn how to work with APIs. Um, and so it just gradually became, um, something that I used more and more.

Blake: Uh, but I think it was a great way to learn because I had very tangible exercises and things that I wanted to do with the tools. I think oftentimes that's where people can get stuck. Um, you, you kind of get into. What I would call almost like a tutorial hell, um, where it's like, okay, [00:03:00] what do I need to learn next?

Blake: All right. Well now what's the next uh tutorial that's online for me to learn and you're not sure where to go I honestly just encourage people to like find something in your everyday life that you want to get rid of that You want to automate and just try to figure out how to do that. It's not gonna look great It's not gonna work great, but it's yours.

Blake: You're proud of it and it gives you something tangible Uh to to work towards um, so that's kind of what that journey looked like there

Mehmet: Absolutely fantastic, Blake. And I think this is a traditional way of, you know, as an entrepreneur, you know, we always, we say we challenge the status quo, right? So you challenge the status quo because you were facing some problems, which is great.

Mehmet: Now tell me more about Shipyard. So You know, I'm sure there was a moment you said, okay, of course you mentioned some of the challenges, you know, with Excel and, you know, managing all the, all the things that Excel cannot do automatically and so on. But like walk me through, you know, the moment where you said, okay, you [00:04:00] know what, I need to have this built into a form of what is today shipyard.

Mehmet: Yeah. And, and how it evolves and. How it moved, you know, from just an idea to where you are today, and I got to talk about later, you know, some great things that you have done. And I was following in the past few weeks, actually.

Blake: Yeah. So the. The biggest unlock for me, uh, was in the early days while I was running the data teams at PMG.

Blake: Um, I was trying to accomplish the same task over and over again for, uh, our clients. Um, so, a great example is, um, just doing something like updating bids and budgets for, uh, their advertising campaigns. Um, I, uh, Replaced the process of using excel for that to where I was downloading the data into the warehouse I was using sql to grab the data and filter it and then I was using python to upload it uh, but now I had another client that wanted to do the same thing [00:05:00] and Well, they had a slightly different data source.

Blake: Um, they had slightly different like R O I goals that they wanted to try and achieve. So how could I build in a template that would do the same process, but would ultimately like change a few variables for their needs and that started scaling across the entire agency. So we were doing it for like 30 to 50 clients, but then we had new use cases that came into place.

Blake: Like what. We had a lot of retailers that had promotional calendars. How can we download the promotional calendars from like smart sheets or Google sheets, and how can we tie that information in them to the products that they sold in their, uh, like SKU, uh, file? And then once we had those tied together, how could we map that to, um, to their, uh, marketing campaigns to be able to run the right ads based on promotions or maybe to turn, uh, ads on and off based on availability of, uh, their inventory.

Blake: And so we just grew more and more use cases [00:06:00] like that. But what I, I found in the space that was really missing was a good tool to templatize those types of solutions. We were writing very scrappy Python at the time. Um, it's not sustainable to have, uh, 30 different versions of the same Python script with slightly, uh, different variables.

Blake: You want to. something that people can plug and play. And so, uh, that's when I started kind of like prototyping, uh, what this could look like, how do I take my Python scripts? How do I let it receive a few variables? But how do I like have like a form UI that someone can use to put in those inputs to be able to run this sort of like workflow, uh, solution.

Blake: What I really saw around the time was talking with a lot of data teams, um, A lot of them are just sitting on data. It's not doing anything. Maybe they have a dashboard on top of it, but it's not actually driving business value. It's not actually like pulling levers and making decisions automatically on their behalf.

Blake: And so ultimately that's kind of what seared me. Towards, uh, trying to, uh, [00:07:00] build out Shipyard. Trying to have a platform that made it easy to templatize the work that you're having to do. Um, uh, to move data where you want it to. To take some action on an external service. And so, in the early stages, it was just that.

Blake: Templated Python that we could run step by step and get logging and monitoring and everything else for. But what we found was really more powerful. Was when we started trying to build out low code templates for, um, other people. So, building out common actions, like downloading and uploading data to literally every single database.

Blake: Or being able to send messages to Slack or Teams. Um, or being able to just trigger external tools like Flutter. Five to load data, DBT to, um, uh, transform data. Um, when we started building out more and more of these templates that were still rooted in Python under the hood, um, that's when it really clicked.

Blake: That's when a lot of people are like, Oh wait, now I understand what I can do. I don't have to write any code. I don't have to do something from scratch. Um, uh, I [00:08:00] can easily. move data between these tools. And that's kind of where we started running with and how we've really grown on our side now.

Mehmet: Yeah, that's fantastic.

Mehmet: And you mentioned something very crucial, I think also, because when, when it comes to dealing with data and, you know, um, everyone talks today about AI as well. Um, so So AI made it easier to achieve, you know, some tasks for automation, uh, like mainly automating the business processes, right? So if you can share like maybe some examples on how AI has transformed the capabilities of, of ShapeYard.

Blake: Specifically for Shipyard, we, we added in some functionality so, uh, that rather than you having to write code from scratch, um, that if you want to deploy like a node that, uh, runs Python, you can use AI to describe exactly what it is you're trying to achieve. So if [00:09:00] you have a data set and you want to like add a new column that, um, uh, summarizes the data over a specific time period or.

Blake: add a new column that has like a specific timestamp or maybe you just want to change the order of the columns or the names of the columns or, um, I keep focusing on columns. I can do way more than that. But, uh, if you write that description, it'll generate the code for you and then you can move on your merry way.

Blake: And I think that's the thing that I am like the most, uh, bullish about, uh, when it comes to AI, uh, is it being kind of like a coding assistant or working directly, uh, Alongside you because I did like a 30 day challenge for myself of only writing code Uh using chat gpt and it did a very good job for a lot of what um a lot of what I needed and this could be various things like on online webinars, uh, taking the HTML of the, uh, like visitors or the registries page and [00:10:00] being able to parse through all of them, uh, to figure out who those attendees are, their name, their title, their company, without having to write a lick of code.

Blake: Um, I recently had to do something where, um, I was, uh, like, trying to take all of the YAML configurations that we generate with our API and then, like, sync these directly to GitHub. Um, wrote most of the code just from going back and forth with, uh, GitHub pilot. Like, it's the best junior engineer that lets me actually focus on the product requirements.

Blake: What is it that I'm trying to achieve? How do I need to do that? And worrying less about the specifics of the syntax and how I need to build it along the way. So that's the area that I'm. super excited on and what we're trying to build more into the shipyard platform.

Mehmet: Yeah. I will go back to the AI just in a moment.

Mehmet: And, you know, one of the things that I start to see you doing also as well, and this is what I was saying, like how, you know, shipyard have, uh, you know, done some great job in [00:11:00] integrating with some known platforms as well. So for example, I saw, I think a couple of weeks back, you did some integration with Databricks, right?

Mehmet: So when it comes to, to integrating with these data analytics platforms and, you know, like these. Data lakes, platforms from, from a business perspective. Like what is the main benefit you see these guys are getting when you give them these kinds of integrations?

Blake: There's really two core benefits. Um, the first one is interoperability between your tools.

Blake: So regardless of what you've chosen to use in your data stack, you can make sure that they, those tools are able to talk to each other in a very simplistic way so that anyone that wants to get access to the data can get it in whatever tool they prefer. Um, the other aspect of it is just like a high level observability into everything that's going on.

Blake: Um, the, the example that I always like to give people is that there's a lot of businesses that do very like schedule based systems to where they might load [00:12:00] the data at 7 a. m., they might transform it at 8 a. m., uh, they might do some processing on it at 9, and then they're refreshing their dashboards at 10.

Blake: Uh, but each of these systems is separate. They're not talking to each other, they just know I run at this very specific time. So what happens if the data takes a little bit longer to load? What if the data loads incorrectly? What if the transformation, uh, messes up, uh, because of that? Um, and so, uh, like all of steps.

Blake: They should be tied together. And you should know that, hey, if the data is taking longer, then we need to delay the transformation step. If it didn't load with the correct data, we need to not run the transformation or refresh the dashboards. And the problem is a lot of businesses don't have that in place.

Blake: So you end up in situations where the dashboard isn't showing the right data. Um, a business user finds it three days later, they ask the data engineering team about it, they weren't aware that it was broken, and then it takes like a 48 hour cycle for people to [00:13:00] figure out what went wrong and how. And so that's ultimately the, the issue that we end up solving for people.

Blake: They can connect each of these steps together. Get immediate alerts. They can be proactive and let like build into the workflow. Something that lets the business users know, Hey, there was an issue with this data and it automatically makes a JIRA ticket with the incorrect data, assigning it to the data engineers to, uh, end up resolving it.

Blake: And so having that level of visibility and making sure that bad data isn't getting deployed, um, that's the other aspect of why. shipyard ends up being so beneficial when you're able to connect all those different tools together.

Mehmet: Great thing. So now you just mentioned before, and you just repeated it now about, you know, sometimes, you know, there's a lot of data that even businesses are not aware they have it.

Mehmet: Right. So, uh, and in, in today's data driven world, so this is something, you know, People think it's a cliche, but it's a fact that everything is data driven today. So what do you [00:14:00] think is the tipping point where data becomes like more of a burden than a benefit? And what do you think, you know, businesses can, you know, use as a strategy or maybe multiple strategies, um, to avoid, you know, being paralyzed by, you know, this huge amount of data that they are sitting on top of?

Blake: I would actually say that, um, the tipping point is almost when you start, uh, like I, I do find that a lot of teams, they focus first on how do we get access to all the data and how do we make all of that data clean and then people will use it, uh, and then they're surprised whenever they finally get it all together two years later and nobody's using it.

Blake: Why can't we get people to use it? Um, I, I personally like taking the strategy of first figuring out, like, what are you going to use the data for, like, working directly with some sort of business stakeholder to figure out, like, [00:15:00] what they're planning on doing once they have this data point, or what are the levers that they're able to pull on their side, even, like, Shadowing them to figure out what like day to day work looks like.

Blake: And then backtracking. Okay. I know this action that I'm going to take. So what data is actually necessary to make sure that this action can, uh, occur? And then you move backwards and figure out, okay, am I able to grab that data from an API? Uh, does it live in some other location? How can I gather it all together?

Blake: Uh, but it ultimately results in you kind of building out the minimum viable data in order to achieve some. Actual, uh, like direct ROI driving activity for the business. And if you do that, you're not going to get overwhelmed because you know that all the data is getting used for specific purposes and you have some sort of visibility, uh, into that.

Blake: Um, that's like one thing that I would mention, uh, to try and make sure that it doesn't, uh, get overwhelming. Another thing that I would mention early on is that a lot of the modern data [00:16:00] platforms, um, or, uh, a lot of the modern data warehouses, they have some sort of logging capability to where you can verify, like, what queries are being run against your warehouse.

Blake: Uh, and this was something that I did, uh, when I was running the data teams at PMG, is verify Every single query happening against that, uh, warehouse so that I could see who is using it, what team is using it, are they using tables related to social data or search data of those tables that I've set up, which ones are queried the most of those tables that are queried, which Columns are they typically relying on?

Blake: All right, those systems, are they powering, um, are they being queried manually in Excel? Are they being refreshed by Tableau? Are they being like run by a script that my team had set up at the time? And so, When you look at the logging data, and you can use that to kind of guide the value of which data is being used the [00:17:00] most, what is it being used for, and what things are kind of underserved, it also kind of gives you an easy way to prioritize, is this valuable?

Blake: Is it worth my time trying to set up and maintain and ensure this data is consistent? Or is it just not getting used at all? And I think there's a lot of people that can't it. answer that question because they didn't start by like use case driven data set up. Um, they just set up a bunch of data. Um, so the, the logging and analyzing those logs can really help there.

Mehmet: So the question here,

Blake: does AI help in this? ? Uh, I think that's a great question. I don't know. Um, I think it could probably look at the, the logs as a whole. Um, I will say that I'm not wholly convinced on AI's ability to, like autonomously an analyze or provide, uh, those results as much as being able to. use natural language to develop the sequel to dig in to the logs and answer the questions that you might have.

Blake: Um, so it could help from that aspect. [00:18:00]

Mehmet: Yeah. Now, another thing, which is you mentioned at the beginning. So when we talk about, um, you know, AI in general and, you know, data and all these things. So some people think, okay, it's just a. Kind of a hype, right? Yeah. So is it really a hype like, or do you think like, no, like moving forward, like this is something tools like chat GPT and co pilot and you know, all the other tools that are coming.

Mehmet: Um, do you think like, um, you know, it's also shape, I would say the way the data team will have to adapt themselves to this new norm, like maybe they would have to learn new skills. Like, what's your take on this?

Blake: So, I think it can be a little bit of both. Uh, it is [00:19:00] overhyped right now. There's a lot of people talking about it consistently and acting like it's the biggest and best thing, uh, ever.

Blake: And in its current stage, it's really cool and can do a lot of interesting things, but I also think generative AI has not quite reached that level yet. quite proven itself of like how it's going to be like best implemented to increase productivity and make things easier. Um, but I, I do think it is going to kind of like change the skill sets.

Blake: Like when it comes to coding, my, my personal belief is that it's going to turn the engineers of the future into just product managers. They are just managing the AI and saying, hear this. Steps that you need to go through. They then QA the output to make sure that it looked correct and that it solves the business need.

Blake: And then maybe there's a system that allows them just the one click deploy that script, uh, to production. And it might not require an engineer. It might just require someone that's, uh, technically minded. That's something that I could easily see, uh, becoming, [00:20:00] uh, commonplace. Um, but what I also kind of see is that, um, Like, with RAG model, so the retrieval augmented generation, are at, basically, when someone asks a query of that, uh, model, you use some search algorithm to look in a database for the right information, surface it to the AI, so that it can have that extra context to hopefully answer the question, uh, more accurately.

Blake: That is really going to require a lot of data work. You have to make sure that the right information exists in that database. Otherwise, yeah, you're, you're going to be surfacing things and helping it with the hallucination, uh, process. But I do think like the natural language interface, um, is going to become more common as we build out more and more AI use cases.

Mehmet: Yeah. You just mentioned something interesting. Also, like, and the question just popped up in my head. About, of course, like you, when, when we deal with the [00:21:00] data. So there is some processes that are needed, which is making sure that this data is clean data. Yeah. Um, so from your perspective, you know, whether with AI, I'm not sure if AI can, can help in cleansing these data, but what do you see, like the best practices to make sure, because I'm, I'm pretty sure like every single organization today, what they are planning to do now, if they didn't start already, okay, we have this data, let's go and train it to the LLM, you know, model, and then try to get some generative AI out of it.

Mehmet: But if the data is. Garbage, as we say, like garbage in, garbage out. So how, how to overcome this, uh, this challenge, I would say, Blake.

Blake: I think people right now with AI and where it gets a little bit of the hype is that they expect it to just be magical and just work. It takes a lot of hard work to make sure that the models can give some sort of, uh, like [00:22:00] accurate response.

Blake: And I think it does. It's very similar to like machine learning back in the, uh, the early days. A lot of the work has to be done up front to make sure that the, the data is labeled properly and that, um, there's not any sort of like. outliers or anything else, because AI is going to have like a general corpus of knowledge.

Blake: Um, it, it has been trained on practically the entire, uh, internet at this point. But when you have very specific domain knowledge, um, that's something where you're going to have to make sure that if you want it to like go through like your, your company documents and like tell someone a specific way that you do business, you're gonna have to make sure those company documents are Uh, up to date, like, um, I, I hear time and time again, that people are just dealing with documentation that either doesn't exist or hasn't been updated in months, uh, despite things changing in the business.

Blake: Uh, and then people like would blame the AI and say, it's hallucinating, it's wrong, it's not giving the right answer. It's like, [00:23:00] yeah, but if you gave an employee access to these training docs and said, okay, like dig in and learn how we do business. And then you ask the same question to them. Would they be able to do the process correctly or answer the question correctly?

Blake: Like, it's just as much a, like, a training data of having the right clean data consistently, um, as much as it is, uh, kind of like a communication problem. Are you clearly describing, uh, the end goal that you want and like, uh, what you want out of AI? And so I do think having clean data is going to become More and more important, but it's still going to be a very like human involved process.

Blake: AI could help you figure out like, Hey, are there any sort of like inconsistencies or incorrectness, uh, in the data that you might not be able to, uh, observe directly. But I also don't think you need AI for some of that stuff. Um, there's plenty of Python packages like great expectations that allow you to test and ensure the data is consistent.

Blake: Um, there's [00:24:00] plenty of like tests you could do with Python to make sure that you're loading things. Um, uh, correctly. And so it's just really dedicating someone to the activity of ensuring accurateness of your data so that you can use AI more effectively.

Mehmet: So I think people need to set the expectation that of course, like we cannot just, as you said, it's not like the, uh, magic pill that we're going to have and then all of our, you know, documentations will be set, everything is, is fine.

Mehmet: And people, they need to, you know, also know that you need to. I would say tie AI and mainly generative AI with other things as well to make it, you know, into action and I would look back, Blake, to the fact of, you know, importance of automation and I'm being, you know, fan of and, you know, evangelizing automation in businesses for quite some time now because I'm saying actually, you know, maybe you.

Mehmet: And correct me if I'm wrong, Blake. [00:25:00] I'm not saying that you don't need AI, but most probably, maybe you need some form of automation first, so you streamline the processes, and then you move forward and see, okay, how I can apply this. Because what I have seen is people, they just want to jump on the one, again, and then, you know, okay, let's do AI.

Mehmet: But actually, they might be able to sort it out in a platform much easier than, you know, applying AI. Do you agree with me on this? Yes.

Blake: I, I definitely do agree with you, and I think the other thing is that, uh, people need to jump back, and rather than just saying, Hey, I have this process, let's automate it.

Blake: Okay, why does this process exist? Does it need to? What's the larger goal that the business is trying to achieve? Can we cut out some of these steps and get it from point A to point Z, but with less steps and make Z, uh, step C instead? So, it's really something where people just need to take a step back and Ask, how can we simplify this?

Blake: And then maybe you can use AI, but you can still probably just write a Python script and [00:26:00] automate something much quicker.

Mehmet: Yeah, indeed. So what are like other technologies that you see, you know, can push what you're trying to do currently with, with shipyard to push it to, to the next level? I mean, of course, like AI is, is obvious one, but do you see like a, Kind of a tying what you do with some other emerging technologies that can really take first the business for you to the next level and also serve your customers in a much better way than you are currently doing.

Blake: I wish I had more emerging, uh, technologies. Um, like I'm actively looking in what's in the AI space. And I, I'm one of the people that like a good old boring old technology is sometimes, uh, like a, a better thing to set a foundation with. Um, but the, the package [00:27:00] that I was mentioning earlier, um, like grid expectations, um, there's, uh, quite a few others in the space, basically anything that can test your data.

Blake: Um, I think that's something that's like severely underserved, uh, in the industry right now. People test their code, people don't test their data to make sure that it looks the same every single day. It has the same columns, it has values in the same ranges, it has only unique values for specific columns, um, but any tools that allow you to test that.

Blake: Whenever you build out workflows that are moving your data or running automation, you shouldn't just set up everything to, to move it to the end point. After every single step, if you have a really resilient pipeline, um, you're going to make sure that, all right, I ingested it. Does it look like I thought it should?

Blake: I transformed it. Does it look like I thought it should? Uh, I wrote some SQL against the warehouse. Does it look like I thought it should? And so every single one of those steps, steps, you're testing the data to make sure that it's accurate. And I know there's some people in this space that are pushing, um, kind of [00:28:00] the idea of what's called data contracts, um, which is like a agreement between the data and the application of how the data should look consistently, what its schema should be, um, like the value types and all that good stuff.

Blake: Um, I haven't dug into it that much, uh, as an idea, but it's kind of the same theme. It's just making sure that your data is. consistent, not just in how it's being delivered and how it's getting to the place, but ensuring that it always looks the same with the same sort of values. And so a focus on that, uh, is going to set you in a really good spot for being able to do more with the data, especially when it comes to the AI space.

Mehmet: Absolutely. Great to hear that Blake. Now I want to shift a little bit gears here and want to ask you, little bit about, you know, being a co founder and CEO, but I can see at the same time, like you pretty much too much into the technical aspect of it. So, so, so how do you balance, you [00:29:00] know, the technical interest that you personally have also as well, because, you know, you started out of a kind of a frustration, like why I cannot automate this, why I cannot do this and, you know, the leadership demands of running Shapr3D.

Mehmet: So how do you do this balance?

Blake: I'll say that, like, every few weeks, I, I have to have some sort of technical thing that I get to dig into, um, whether it's a personal thing or it's for work specifically, um, I would miss it if I didn't get to do any sort of, uh, technical work, but in general, um, a lot of it is just, you know, I'm just delegating it over to my co founder and the rest of the engineering team, um, to, uh, to ensure that the really heavy lifting, um, gets done on their side.

Blake: Um, I can think through, like, the problems of, okay, here's what I want to solve, here's all the steps, here's the executional pathway, and that's, like, where I should stop and, uh, hand it off. Give a few ideas and make sure that others can work through it, because I will say, I can write some mean scripts. I can work with data.

Blake: [00:30:00] I can't build an application. That's what my co founder Eric is for. Uh, that's what the engineering team is for. And so I have to recognize, like, where I'm not, uh, as, uh, well served and how I can make sure that I'm hiring the right people on the team to be able to dig into these problems, uh, deeper along the way.

Blake: Um, ultimately, when it comes to running the business, it's just about figuring out the vision. And figuring out how you can align the right people to hopefully make that vision a reality.

Mehmet: That's absolutely great to hear also Blake. Now, from, from, you know, again, business perspective and growing the business.

Mehmet: And I just mentioned, you know, a couple of, um, you know, kind of, uh, platform you integrated with. So this kinds of. You know, let's call them partnership. If even if they are not like a direct partnership, but kind of integrate through the APIs and so on. So how this helped you, you know, to scale the business?

Mehmet: Because, you know, when I talk [00:31:00] to some other founders, I always tell them, like, try actually To build these partnerships to scale your business. So how crucial was that for you, Blake with Shipyard?

Blake: I will say that early on we focused a lot on building out integrations with most tools in the data stack But we didn't do as great of a job of reaching out to that team and figuring out how we could work Hand in hand with co marketing and figure out how Um, our customers could work better together.

Blake: Um, that's something that we're getting much better at right now. Um, with, uh, partners like Databricks and Coalesce, um, and, uh, trying to make sure that we're able to tell that story better. But I will say in the early days, like that was the big unlock for us. Um, integrations are a little bit of a race to the bottom because as soon as I have 50 integrations, someone comes in the door and they're like, Hey, do you have something for this tool?

Blake: And the answer is no, but we can build it. Um, and so, uh, you just, you're constantly making more and more new integrations, but I think you [00:32:00] eventually get to a point where people, like, realize on the tool that, like, hey, pretty much every single tool that I use, uh, right now, um, I can Run some sort of action against it.

Blake: I can connect it together. And I think what's really helped us is having that level of transparency into how we build integrations. Um, as far as I'm aware, we're one of the only tools in this space that takes integrations and open sources them as well. So you can see all the Python that's running under the hood that we've built.

Blake: Uh, you can inspect it. You can fork the code and use that code on your own, um, directly within the platform. Um, but that was super important for us to make sure that we could give that level of transparency, uh, into everything that's going on. And I think that really helps stuff as well. Um, it wasn't just a proprietary integration that you wouldn't know how, um, How it's achieving, uh, things under the hood that your engineering team would eventually have to rebuild.

Blake: No, it's something where you can guarantee that it's [00:33:00] secure, that it's running quick, it's using the latest and greatest techniques, uh, and we've been able to even tighten up those integrations more by working directly with the partners and having, um, um, having them kind of evaluate the code themselves and provide like guidance and feedback.

Blake: And so it's helped us on that, uh, that front as well. But I will in general say that it's just. It helps you scale. Being able to have integrations means that you're able to service

Mehmet: more customers. Absolutely. Absolutely. So, uh, you know, like, um, I'm really impressed, you know, but by what you did so far, uh, they can really, you know, I can see, you know, you're, you're moving.

Mehmet: And I know like recently also you had your, uh, SOC 2 compliance also as well for, for the platform. So what's next for, for shipyard?

Blake: Over the course of the next year, a lot of it is just battening down the hatches, uh, and making sure that we have, uh, all the best, uh, like, [00:34:00] enterprise level features, uh, along the way, um, and focusing a lot on security.

Blake: Because ultimately, we're dealing with, uh, a lot of companies that are in highly regulated industries, and we're having to make sure that we have everything, uh, that those companies, uh, need to service them super effectively. But I'd say after that. The big thing that we're focused on is, um, continuing to improve the ease of use, getting to a point where people can build step by step in the platform and validate, uh, like the output, um, after every single step of the workflow, but then incorporating that with AI, because I envision a future where Where we talked about having the product manager, uh, over the AI, that I can build a step, describe what I want, see how it looks correctly, tie that, uh, the output of that step to another step, and build it that way.

Blake: And so, kind of having an interface where people can. Right now, it's drag and drop and it's YAML on the back end, but I want it to be something where you can have a conversation, uh, to build that step, but it's still [00:35:00] code on the back end. So it's conversations as code and building the workflow hand in hand that way.

Blake: I think that's going to be the future and so that's ultimately what we'll be steering towards afterwards.

Mehmet: That's great and I'm looking forward to see that in action because this also will enable I would say maybe non technical teams, or let's, let me put it this way, maybe smaller organizations that they might not have, uh, the technical talents yet in house.

Mehmet: If they don't

Blake: have the data team, it will be very helpful.

Mehmet: Exactly. So this will open doors for a lot of, uh, uh, of benefits for, for such companies. Now as we come almost to an end, Blake, so this is a question I always ask to my guests at the end. So kind of a. words of wisdom, advice to fellow, um, people maybe in, in data science, or maybe founders who are on the verge to enter this domain of, of data and AI.

Mehmet: So what, what kind of one piece of advice you would give them [00:36:00] and where to find more about ChipYard and about you?

Blake: For people that are early on in the data journey, um, my biggest piece of advice is don't worry about all the technical stuff. Don't worry about all the tools. Just figure out what is a problem that the business has that can drive, uh, revenue.

Blake: And then how do you grab the data in order to be able to solve that? If you focus on problems versus focusing on like, clean data sets, you'll be at like the top 10 percent of all data teams and be able to drive more value and stop being a cost center. You can start actually being like a value driving center, um, and kind of grow your career in your business, uh, that way.

Blake: So that'd be my biggest piece of advice. And if you're looking to, um, find more about shipyard, um, you can find Find us online, shipyardapp. com. Um, I'm on LinkedIn, uh, pretty regularly. Uh, and we also have a weekly newsletter where you can stay up to date on the latest in the data space. It's called all hands on data.

Blake: So feel free to subscribe to that as well.

Mehmet: [00:37:00] Sure. I will make sure all the links will be in the show notes. So people, they don't have to search that you will find the links. Uh for the website and for the newsletter in the show notes Blake really I enjoyed the conversation with you today like very insightful very Engaging also as well and i'd love to hear the stories Behind, you know shipyard and how we started it and also all the great work that you're doing currently So thank you again for being a guest with me today.

Mehmet: And this is how I end my episodes Usually this is for the audience if you're first time You Listener or you're seeing us now on youtube. Thank you for passing by if you liked what you heard Please subscribe to our podcast. We're available on all podcasting platforms And share that with your friends and colleagues and if you are one of the people who keep coming Thank you for being loyal to the show.

Mehmet: I really appreciate that. I appreciate your feedbacks. Keep them coming You know where to find me. I'm more active on linkedin and if you are interested to be on the podcast Also, don't [00:38:00] hesitate to reach out to me. I would be more than happy to discuss how we can arrange that Thank you very much for tuning in and we'll meet again very soon