Status Go: Ep. 226 – AI – The Continuous Emerging Trend | Jake Miller

Episode Highlights

[00:00:00]: Measuring Productivity

[00:00:22]: Intro and Welcome

[00:03:03]: Introducing Jake Miller

[00:06:09]: The Birth of Engineered Innovation Group

[00:08:38]: AI – The Emerging Trend for Years to Come

[00:11:43]: Using AI Day-to-Day

[00:15:43]: Productivity Gains in Development

[00:19:28]: Improved Quality as a Byproduct

[00:23:06]: Teach More than Coding

[00:27:55]: AI Impacts on the Developer Career

[00:29:33]: Crystal Ball 3 Months and 3 Years

[00:32:41]: Thank You and Close

 

 

Episode Transcript

 

Jake Miller [00:00:00]:

First and foremost, we have to be able to have a baseline to measure productivity, right? What I am having a hard time doing is quantifying that right now, and I think that is a nut we are going to crack here soon with a lot of this process automation that we’re putting in place by using LLMs and AI.

Voice Over – Ben Miller [00:00:22]:

Technology is transforming how we think, how we lead, and how we win. From InterVision, this is Status Go, the show helping IT leaders move beyond the status quo, master their craft, and propel their IT vision.

Jeff Ton [00:00:43]:

Welcome to status. Go. I’m your host, Jeff Ton. Before we get into our episode today, I have a favor to ask. If you enjoy the Status Go podcast, whether you are a frequent listener or this is your first time joining us, tell a friend, send them a link, or post on social media. Let others know about us. We appreciate it.

Now, as they say…on with the show.

One of the things that I love about hosting this podcast is the discussions we have about emerging trends. We usually do an episode or two at the beginning of the year to talk about the trends you should be keeping your eye on. One of those trends that always seem to be out there in the future someplace is AI. We’ve been talking about machine learning and AI for several years now, but always in that futuristic mindset, or maybe only for those companies with large data science budgets. But all that changed with the launch of Chat, GPT, and other generative AI models.

It just exploded on the scene earlier this year, and now companies of all sizes are wrestling with what do I do with it? How do I use it? What do we do? What constraints do I put on my team? It’s a conundrum that many of us are facing.

AI is such a game changer, and it’s evolving so quickly that we’re likely to be talking about it as an emerging trend for years to come. Today’s guest is Jake Miller. Jake is the CEO of the Engineered Innovation Group. EIG’s mission is to help organizations build new and meaningful useful products and services. Jake is passionate about AI and its impact on innovation. I’ve had the opportunity to listen to him several times and host a panel that he sat on as well. And I guarantee you, you will walk away with insights, and you will walk away energized by the possibility this technology can open up for you and your organization.

Without further ado, Jake, welcome to Status Go.

Jake Miller [00:03:00]:

Thank you so much, Jeff. I’m really excited to be here and for the conversation today.

Jeff Ton [00:03:03]:

Jake, I mean it sincerely. I have enjoyed every one of the conversations that we’ve had the opportunity to have over the last several months, including sitting across the table at dinner at an Indiana CIO network event where you were on the panel. Before we dive into this topic of AI and what you’re seeing and how you’re using it, can you share a little bit about your background, your career journey, and what led you to the founding of EIG?

Jake Miller [00:03:34]:

Yeah, I would love to share. So my background is pretty much in software engineering, though, right? Before I became a software engineer, I studied English at university. Oh, wow. I was writing code, actually, in high school. Surprisingly, I learned some COBOL.

Jeff Ton [00:03:55]:

All right. COBOL is still alive and well!

Jake Miller [00:03:59]:

Apparently! And I went to university. So, I went to IU at the Indianapolis campus, and I said, you know, I want to…I’m having fun with the computer science program, the computer program here, but I kind of want to do something that will make me a little bit more well-rounded, because I was already writing code. I was actually already making money from writing code as an independent consultant in the days of, uh, you could actually make pretty good money.

And I said, well, let’s study English. So, I did, and I found a love for linguistics, actually, at the time. And so I really got deep into that. And then, having a computer and a technology interest already, natural language processing became something that was really fascinating to me. Yeah, didn’t do a whole lot with that except put it in the back of my mind for the next 20 years.

So, after graduating college, I continued writing code. Then about two years in, I started getting into engineering and product leadership. So I’ve worked for companies in the healthcare industry, like Indiana University Health. I worked for Allison Transmissions as a contractor on some really cool projects. Then I went to a company called Exact Target…that folks that are in Indianapolis are probably very aware of, and subsequently was purchased by Salesforce.

I was director of engineering there for the Marketing Automation Suite of products. After that, I did a startup, co-founded a startup and served as CTO there. And then after about four [years], you know, I’m ready to go do my next thing. I am a creator at heart. I still write code to this day. Not as much as I would love to. I still work with lots of startups. In fact, I’ll talk about what EIG does here in a second, but I wanted to do that on a daily basis. I want to take ideas and bring them to life. Not just like bring an idea to life and then stay with it for 20 years. I wanted to really just help people bring their visions into the world. And so that’s kind of how I got to where I am today.

And I think you asked about how did EIG get started?

Jeff Ton [00:06:09]:

Yeah. Tell us a little bit about that story, and then give a little bit more color as to what EIG does for its clients.

Jake Miller [00:06:17]:

Yeah. So, the Engineered Innovation Group, you can think of us as a software company, so we do software development for sure. But I don’t like calling us a dev shop. I like calling us an innovation agency. And the reason is we don’t just build software products, we build software organizations. So, we provide product design, product marketing, software engineering, of course, artificial intelligence and data science services, site reliability, quality assurance.

We will build out all the process, all the security and compliance needs, including the product itself, and help our clients build their organization. And eventually, they are typically a VC-backed startup. They’ll want us to peel off, and keep their own people, but they have a fully functioning software company now.

We do the same thing, a slightly different flavor for mid-sized businesses and enterprises. But that was…the genesis of the company was – let’s take the brain or the cognitive overhead that every founder faces when they’re trying to build a software company. Take that out of the mix. Everyone needs to be able to log in. Everyone needs to have security, compliance, and privacy. Everyone needs infrastructure. Why are founders focusing on those things? Focus on your differentiated innovation. There’s 10% of the thing out of all that that you really care about, and that’s your domain of expertise. And I wanted people to be able to do that. So that’s what we do.

The concept came out of my experience working with various startups. Specifically, when I was with MetaCX…High Alpha, we were a High Alpha portfolio company, and I was working with other CTOs and founders, and I just saw the same mistakes being made over and over again. I also saw the timeline, and this is what I’m about to say is true for all venture studios. The timeline, from an idea, getting a green light in an investment, to finding a founder, a technical founder, building a product, and getting something in hands and in market is a six to twelve-month process. Why in the world is that not a three-month process when 80% is rinse and repeat for the most part, right? I’m generalizing, so that’s where it came from. And over the past two years, we’ve just added service by service until we were able to have that full suite to provide to clients.

Jeff Ton [00:08:38]:

And for our listeners, especially those around the Indianapolis area. It seems like every other day, I’m getting an update on LinkedIn that one of the area rock stars has joined EIG. You are building an amazing organization. So, shout out to Karen and Hutch. I know they are on board with you folks, and I just can’t wait to see what all you’re doing.

But let’s dive into this. In the opening, I talked about AI as an emerging trend for years to come, and I actually stole that from you when we were talking a few days ago. What things are you seeing that lead you to that belief?

Jake Miller [00:09:27]:

Yeah, so I actually like to look at macro trends in technology as my backing data for this. The major one is in the amount of data that companies have been storing. So, the 90s were about, oh, we have our application database, and now we need to get analytics from the data in that application. Out come data warehouses, people that are managing those, they still exist, of course, but then that moved on to, oh, well, now we have every person in our organization and all of our customers and all of our devices are producing information. We don’t know what to do with it, but we’re going to go store it in a data lake.

And so, you amass all of this data, then come the neural networks, really start getting the light of day, and we start realizing, wow, there’s data science that we can apply here. There’s machine learning techniques that are starting to pop up their head. But we have all the ingredients now for this to really become exponential.

The technology applications become exponential, meaning speech and audio, like speech recognition, text-to-speech, emotion detection, the vision technologies for recognizing objects and people and faces. Generative AI, the hot topic right now. Creating text, summarizing documents, generating images, creating films, all those things.

When you look at each of those subdomains of machine learning and artificial intelligence, have matured to a point that now we can start connecting them. And when you start connecting them, for example, if we wanted to create a…and you actually had talked about a podcast that you did where you were using a large language model to interview, right?

Jeff Ton [00:11:20]:

Yes.

Jake Miller [00:11:20]:

Imagine putting a person’s face…personifying that a digital human twin with that model behind it, and that sophistication growing over time. Imagine the possibilities. We’re going to be talking on our next podcast, Jeff, I hope, where we have one of my digital colleagues with.

Jeff Ton [00:11:39]:

We’re going to be talking on our next podcast about all that!

Jake Miller [00:11:39]:

We’ll do that. That’s where…where we’re going and

Jeff Ton [00:11:43]:

What I mean is, well, your excitement about this comes through, and I know a large part of this ties back to your background in linguistics and these large language models and natural language processing. It is all coming together. One of the things that I have found in our conversations, Jake, is that you are actually using AI at EIG. Not only are you helping your customers explore it, but you’re using it. How have you brought this in as a tool to leverage for the work that you do?

Jake Miller [00:12:24]:

Yeah, I love this question because we’re finally getting to a point where we’re starting to see traction. And some of these won’t be a surprise to people, but one of them might be. So first, we started just experimenting like every other company on the planet, right? Like, yeah, large language models, what can we do with it? We’re keeping privacy in mind, especially being an agency. If we’re working with clients, if we have data or code from them, we don’t put that into a large language model, at least not without permission.

Jeff Ton [00:12:56]:

Right.

Jake Miller [00:12:56]:

So, we’re using ways like Copilot, so code, generation code, assisting, that sort of thing. Not a surprise, probably at all. We are using it to help generate marketing concepts. So not actually building marketing, but more as a sounding board. And I’ve actually found personally, this is something that’s been very helpful to me. And actually, I’ve had a couple of engineers when I poled the whole company, I’m like, hey, how are you personally using LLMs in your day-to-day work that we may not know? And two individuals said the same thing that I had been doing. I use it as a sounding board. I have an idea, I throw it out there, and it kind of throws back some other ideas, and I go, oh, I wouldn’t have thought about that.

So, it’s like a companion. Pretty cool. I think a lot of people are doing that. The other, I think, is where we start seeing more sophistication and true application coming to flight. And we are starting to use LLMs with a couple of our internal projects as a pilot to take the information of people communicating in Jira and all the information about the code, all the information about requirements, acceptance criteria, what’s working, who’s doing what, what’s the status of things. So, you can say, or I can say in the morning, Hey LLM, what is Joe doing? Or what has Joe done last week on such and such in the project? And it can summarize it for me, like being able to consolidate real-world information quickly is hugely valuable to anyone, really, but especially very busy executives.

Jeff Ton [00:14:34]:

Well, one of the comments that you made is you said this a couple of times, actually, just like everybody is doing, I think you’re selling yourself a little short. I think you are leading the charge on some of this, Jake, because I’ve talked, at least in the corporate technology world, not necessarily in the technology product technology service world, in the corporate technology world, I think they’re taking a little bit slower approach. So, I think the lessons that you are getting from this are going to be helpful across the board.

Now, I’m going to pause right there. We’re going to listen to an ad from InterVision Systems. As our listeners know, InterVision is the publisher of the Status Go podcast.

Voice Over – Ben Miller [00:15:28]:

Unlock the Power of More. With InterVision Systems, we provide the cutting-edge technology and expert guidance you need to take your business to the next level. Don’t settle for less. Choose InterVision Systems and discover what’s possible. Contact us now to learn more.

Jeff Ton [00:15:43]:

And if you want to learn more, visit InterVision.com.

Today, we’re talking with Jake Miller of EIG, and we’re talking about this emerging trend, this exploding trend of AI, and how Jake and others at EIG are using it to advance their work and advance their clients’ work.

I heard you speak…I think this was the panel that I moderated, Jake. So, a couple of weeks ago you mentioned productivity gains from your development staff, from your engineering staff, and before break you mentioned that you’re using Copilot. So, what kind of productivity gains are you seeing first of all? And then we’ll dive a little bit deeper into that.

Jake Miller [00:16:37]:

Yeah, I’m glad you asked this question because it’s something that I’m very passionate about. One of the ways that, first and foremost, we have to be able to have a baseline to measure productivity, right? Yeah. And as a new agency, this is something that we’ve continued to work on. And I was actually talking to actually Susan Orr at the IBJ the other day about this exact topic, and she asked me the same question. I said, well, you know, it turns out getting my baseline has been a little bit more challenging than I thought. So that doesn’t mean we’re not going after it, but we are.

So, the numbers I’ve been saying, I want to see a 30% productivity increase in my organization over the next year leveraging AI. And that is still a true statement. What I am having a hard time doing is quantifying that right now. And I think that is a nut we’re going to crack here soon with a lot of this process automation that we’re putting in place by using LLMs and AI.

Actually, some of the things I actually just talked about, like summarization, auto-creation of acceptance criteria, that sort of stuff, right? It’s kind of hard to quantify, frankly. It’s kind of a qualitative measurement. But I can give you categories of things that we’re using that I am seeing anecdotal evidence that we are improving.

So, the first is Copilot itself, right? The ability for us to increase individual productivity. We are seeing categories, just communication. Specifically. I just talked about, like, Jira. So Summarization, we have now automations in place that are doing things like checking if someone’s entered a ticket for work. And the description, it’s not just checking, does this description exist? It’s saying, is it quality? So we’re able to take that information from that ticket and say, let’s go ask it to be summarized by an LLM and use a rubric to see if it is a quality input or not. And if not, we’re going to publicly shame that person now…just kidding! We’re going to notify someone that there is a ticket put into a work status that really is not intelligible. At least flag it. Again, it’s an experiment, but I’ve already noticed it increasing quality of the work we’re doing. So that’s another bucket.

And I would also say just things like writing copy. Not going to lie. I think at least getting frameworks in place for presentations and such. It’s been very helpful on that front.

Jeff Ton [00:19:28]:

Yeah, I use it for the podcast. I use a tool called CastMagic. And what that platform does is it puts a UI on the front of, I think, Chat GPT on the back end. But it ingests the audio from the podcast, and it creates marketing collateral. I mean, it’s not finished product, right? But it’ll go through, and you’ll see the result of this when I send it to you later, Jake. It will go through our discussion today, and it will highlight quotes from you that it felt were important for our conversation that you can then use in marketing. And it does a bunch of other things, but saves four or 5 hours a week.

But I love this concept of leveraging it for developers. I interviewed a gentleman yesterday for the podcast that’s coming upcoming, and we talked about the developer experience. We’ve all talked about customer experience, employee experience, but developer experience and how tools like AI can automate some of these things and make the developer’s life easier so that they’re more focused on the things that are vitally important to the end product. Are you seeing those results? Also? You mentioned quality, but are you seeing that they’re able to focus more on some of these other tasks now?

Jake Miller [00:21:00]:

I don’t think we have been able to fully realize its potential yet. I think it’s just time will tell. I feel pretty darn confident that by the end of this year, we will start realizing that. I think one of the problems, again being our company is almost two years old and we’ve really only had a team for a year other than myself and another person. And so, what we’re rolling out with process-wise and such, I think that’s the stuff that developers don’t like doing, and that’s where I want us to focus. Productivity gains.

While I think it’s really interesting, like the coding side and productivity gains from writing code, I think there’s definitely use cases there. And actually, I’ll give you a real-world example here in a second. I think we can’t forget in the realm of developer experience, what’s all that other stuff that takes one or 2 hours a day? Documentation, making sure your statuses are in the right.

Jeff Ton [00:21:59]:

Every developer loves documentation.

Jake Miller [00:22:01]:

They love it.

And it’s something I get really excited about. And we’re doing this more on the customer front because APIs we develop aren’t our APIs, it’s our customers’ product, right?  But talk about developer experience. Right now, developers have to go to a website and look at the API. And you guys probably talked about the last podcast, but look at the API documentation, understand what every field is doing. Yada yada yada.

There need to be tools in place where I say, I’m trying to connect this thing here to that thing there. They both have APIs. Write the code to do it, and it gets you 70% of the way there. And I also think it’s such an important point for people to realize: don’t think about these tools as replacements; think of them as augmentation. And what I mean by that is if you can get even 50% of the way where you need to go as a developer by using a tool. Wow, that’s absolutely.

Jeff Ton [00:23:06]:

I think that’s great. And I’ll give a shout-out to Okteto. It’s a developer experience platform. And I interviewed the CEO yesterday, Romero, so I love this because my background is you talked about Cobol in your opening. I was a Cobol programmer. That’s what I did, man. We didn’t call ourselves developers or engineers then. We called ourselves programmers. But I have this warm spot in my heart for the development team, and it’s about time that we invested in tools for them.

Now, I want to go back to another comment that you made as you were on a panel, and I know you probably love that I keep bringing this back up again.

Jake Miller [00:23:56]:

Every time.

Jeff Ton [00:23:59]:

You were on a panel at the Indiana CIO network, a gathering of CIOs, CTOs and other technology leaders, and you said something that really caught my attention. and it was basically a message for the coding academies that are doing a fantastic job of bringing coders to the marketplace. But you talked about the future and what they need to be paying attention to. Can you share that message and elaborate a little bit for us?

Jake Miller [00:24:33]:

Yes, I can.

So, there are two parts to this. One. One’s sort of an immediate, and the other one’s a little bit longer term. And the first one is, and I don’t think I said this in that particular interview, but the first one is we’ve got to start diversifying the content for students coming out of coding academies. I interview so many people from coding academies, and I cannot tell you how many are taught React, JS, all of them. And that’s it. Please, let’s start teaching computer science, fundamentals…data structures, algorithms, and please, let’s start teaching data.

And when you start seeing these trends in AI like we’re talking about, and this will segue into the part two of this, is we need more people that understand how data works, how data is stored, how data is processed, how data is structured. And that’s a massive gap. And I’ve seen that for five, six, seven years, for sure. I’m hoping that we start seeing that gap get filled because it’s a big opportunity, not just for the coding academies but for the people that are going through those academies and looking for the job opportunities.

That’s a big that’s step one. Step two is what is going to happen with software engineering jobs. So, the hypothesis, like any intelligent person, I keep changing my perspective on this. We’ll see if I gather information…

Jeff Ton [00:26:06]:

You’re a large language model, and you’re learning from the data.

Jake Miller [00:26:12]:

That’s exactly right. That’s exactly right.

But the question comes up frequently: what’s going to happen to software engineers? Will their job be replaced? And if they’re not going to be replaced, what’s going to happen? And I foresee what’s going to happen is things that are higher level programming. That’s not a judgment call on whether it’s a good or a bad. That’s like higher level.

So, developing frameworks for an interface. Let’s say you’re building a web application, and you need some scaffolding for code that’s going to be automated. You’re going to talk to an LLM, or you’re going to or and you can actually do this today. There’s some tools, rudimentary fundamental tools, but there are tools where you can provide it a design, and it will go write the code for you. There’s going to be more and more of that. So, what does that mean? Well, those are the things typically we assign to the more junior engineers.

Jeff Ton [00:27:10]:

Yeah.

Jake Miller [00:27:11]:

So that means you’re going to have to have the folks that are more specialized or that have the more in-depth knowledge of the technology, like how data works, are going to be the more sought-after individuals. Just a hypothesis, but I see us going that direction.

And, the sophistication of technology is going to require when we’re working with LLMs, or not just LLMs, but I will use artificial intelligence as the broad umbrella. We’re going to need more people that know how to train the LLMs, to train the technology to do what we want it to do versus smashing keys on a keyboard and typing, know, discrete lines of code.

Jeff Ton [00:27:55]:

I had a conversation the other day with John Light, the CEO of Sabertooth out of the Houston area. And he was talking about, and I kind of agree with this as well, is that in addition to the points that you just made, we need people to understand more about the data, more about the large language models that we’re going to have, almost a verticalization or a specialization. You’re not going to be able, as a developer, to float between retail and manufacturing, manufacturing and healthcare,  healthcare and another industry. You’re going to have to have so much domain knowledge to marry with the…because data behaves differently in different domains, right?

Jake Miller [00:28:45]:

Yeah.

Jeff Ton [00:28:46]:

So, you’re going to have to have that as well. Are you seeing that as a possible trend as well?

Jake Miller [00:28:59]:

To play devil’s advocate on that, I would be interested to see if the models that we’re creating will become those domain experts. In fact, it would be the opposite; generalists of the domain have more access to opportunity because we’re augmented with those experts.

Jeff Ton [00:29:33]:

Well, Jake, I want you to get out your crystal ball here, and I’m going to ask you to predict the future in two different time frames. So, we’re sitting here as you and I are talking. It’s August 2023. What do you think we’re going to be seeing out of these LLMs and the incorporation of them into our businesses? What do you think we’re going to see in the next few months? Let’s call it six months. And then the follow-up to that would be, how about a year or two down the road? So, six months and then a year or two down the road?

Jake Miller [00:30:16]:

Yeah. I love it. So, I think the next three to six months, what we are going to see are real-world practical applications of LLMs. There’s been so much experimentation, and part of my job is I see startup ideas and concepts come across my desk every single day, and 95% in the past three months have been with LLMs and we’re going to see what’s working and what’s not. So, just like I talked about earlier, I had a prediction two months ago about what my company was going to do. Turns out it’s not exactly what I thought it was going to…we’ll get there, but everyone’s going to start having those realizations about what’s working for them and what’s not as an organization.

I think more broadly speaking, we’re going to see more security and privacy tools that are empowered by AI. And I don’t just mean like for network monitoring or intrusion detection, I mean tools that are helping us be smart about privacy information and IP going into LLMs and that sort of thing. I think that is going to open the door majorly for big corporations to start being comfortable, opening up opportunity to their employee base, where then we’ll get to see what’s the next three months really do. Yeah, what it can really do in those bigger organizations. So that’s kind of how I see this progressing. Do you want me to jump into my long-term too?

Jeff Ton [00:31:42]:

Yeah, absolutely.

Jake Miller [00:31:45]:

I’m going to say my three to four-year prediction is the human-computer interaction that you and I know today. That is the mouse and keyboard. Primarily. Some voice is going to be displaced by some percentage. I’m going to make one up: let’s say 30 or 40% by pure human-to-digital human interaction. And so, I’m not going to be typing, I’m not going to be using my mouse to point and click. I’m not going to write code. I’m going to say, “Hey computer,” or “Hey, Paula, I’m trying to figure out x, y and z of this information and this file. Can you give me the forecast of blah, blah, blah?” “Yes, I can do that. Here’s the result.”

That is where I think we are going, and I will think it is a big miss on our part as human technologists and innovationists if we don’t do that.

Jeff Ton [00:32:41]:

Yeah, I look forward to that day in some ways. And in other ways I’m like, oh my God, is the person sitting next to me on the plane going to be talking to their AI model while we’re trying to fly? And it’s no, no, but no. I love those predictions, Jake, and would love to have you back maybe in the middle of next year or something, and let’s carve that up and say, well, where are we in that? Right? Because the beauty of it is it is emerging, and it is changing over time.

Jake, I have to thank you so much for being on Status Go. I learn so much every time you and I talk. I enjoy our conversations. I am already looking forward to the next time we get a chat. So, thank you for carving out time for us today.

Jake Miller [00:33:31]:

Absolutely, the same to you. I always enjoy our conversations. I appreciate it.

Jeff Ton [00:33:35]:

To our listeners, if you have a question or want to learn more, visit InterVision.com. The show notes will provide links and contact information. And if you’re interested in continuing this discussion, look for The Status Go podcast group on LinkedIn. We’ll get a dialogue going about AI and where you’re using it in your organization.

This is Jeff Ton for Jake Miller. Thank you very much for listening.

Voice Over – Ben Miller [00:34:07]:

You’ve been listening to the Status Go podcast. You can subscribe on iTunes or get more information at intervision.com. If you’d like to contribute to the conversation, find InterVision on Facebook, LinkedIn or Twitter. Thank you.