Where AI Is Actually Working, with James Barnes
In this episode
- 00:51 Jon’s path into software and aviation: Boeing, the aftermarket, and the pull back to parts
- 04:03 The problem: why parts procurement stayed manual and fragmented
- 06:35 Why nobody solved it, and the Amazon effect aviation never got
- 10:52 The pivot: from a marketplace to an agentic platform
- 20:07 Building AI with AI: the first session in Claude Code
- 24:22 Inside the sourcing agent: the data lake, confidence scores, and humans in the loop
- 33:51 The AI foundry: small teams, high output, governance kept human
Read the full transcript
Click hereJames Barnes: That’s I think the most important story to get out there today is is where AI does solve problems. Right? Where it’s not speculative.
Benjamin Tosado: Yep. Yep. Nope. Completely agree. So let’s talk about the industries a little bit.
So in terms of aviation, I mean, what what are you seeing out there, Barnes? Like, what what types of solutions do you think are really interesting, AI solutions that are getting deployed in aviation today?
James Barnes: Yeah. Aviation is interesting because it is it is both compliant and compliant adjacent.
Benjamin Tosado: Welcome to Cleared for Takeoff, Real AI stories from experts in industries that move the world. I’m your host, Ben Tosado, founder of Deep Blue Cloud. We’re launching weekly episodes on Agentic AI and what it means for aerospace and defense. Hit follow so you never miss an episode. Today, I’m sitting down with James Barnes, our VP of operations for our very first episode of the Cleared for Takeoff podcast.
Hey, James. How are you?
James Barnes: I’m doing good, Ben. The question is, is this episode zero or episode one?
Benjamin Tosado: I guess, you know, I guess this would be episode zero. Right? This is our test run. We’re kinda we’re kinda figuring everything out. Super excited to start the project with you, though.
James Barnes: Yeah. Absolutely. Let’s get in.
Benjamin Tosado: Cool. Why don’t you, tell tell the listeners a little bit about yourself, your background, how you became to be at Deep Blue?
James Barnes: Yeah. So I’ve, I stumbled into AI, well, not AI. I stumbled into IT as a second career essentially. So, my my IT journey was bottom up, right? So, I started bottom up, you know, doing audio visual, doing video conferencing, and then that led me into a whole world of of, you know, vertical promotions.
Got all the way promoted up to essentially enterprise architect level. And that was all in a highly regulated government industries. So from there, I pivoted, moved into commercial landscapes, got into cloud computing, AI, and joined Deep Blue in roughly around 2023 versus an advisory position and then came on full time as a VP of operations. And and my scope really is client delivery. So all aspects of client delivery and and operations.
Benjamin Tosado: Awesome. So you’re you’re actually seeing where, where we’re helping a lot of these customers actually implement solutions that we’re gonna be talking about on the show. Absolutely. That’s cool.
James Barnes: Yeah. And cradle to grave too. Right? So from initial scoping all the way through to delivery and then, you know, post delivery. Right?
And how how can we continue to help customers evolve and adopt?
Benjamin Tosado: Yep. Yep. Absolutely. So let’s talk a little bit about the show. Right?
Why we decided to start, Cleared for Takeoff. My my vision here was I mean, I think we’re doing really cool things with customers that are in some of these industries. You mentioned highly regulated industries. We’re doing some stuff there like in aviation and the manufacturing, defense, industrial base. We’re also doing some stuff in distribution and doing really kind of cool things around AI in all of those industries.
So the goal here is to talk about some of those cool stories and bring guests on that will be able to talk about some of the work we’re doing, some of the things that they’re seeing in the industry, and to help our listeners understand, how AI is actually getting applied on the front lines in, aviation manufacturing and distribution.
James Barnes: Yeah, absolutely. And I think, you know, what’s really important is practical application. Right? Because we’re we’re we’re cutting edge right now with AI. It’s moving faster than than people can report on it.
Right? So how how are people using it effectively without, you know, without it being pure theory, without it being pure speculation? How are they solving real problems? That’s I think the most important story to get out there today is, is where AI does solve problems, right? Where it’s not speculative.
Benjamin Tosado: Yep, yep. Nope, completely agree. So let’s talk about the industries a little bit. So in terms of aviation, I mean, what what what are you seeing out there, Barnes? Like, what what types of solutions do you think are really interesting AI solutions that are getting deployed in aviation today?
James Barnes: Yeah. Aviation is interesting because it is it is both compliant and compliant adjacent. Right? There’s a massive amount of controls for very good reasons. Right?
So if you talk about, you know, parts procurement, if we talk about line maintenance, if we talk about, you know, sourcing parts and maintaining inventory, right, there there are all of these elements that are highly human focused traditionally. It had massive amounts of impact on the bottom line. So applying efficiencies using AI processes to augment what people do to make them do something more efficient or to apply more standardization to to processes. It all impacts things like safety. It impacts, you know, revenue, right, and and profitability.
So it’s it’s intensely important in a in an industry like this to both apply AI to solve these problems while maintaining these compliant postures. And then really drive outcomes that can be tangible, can be measured, and can be eventually approved, improved upon.
Benjamin Tosado: Yeah. No. Yeah, I think that’s that’s absolutely right. And one of the interesting things that you said that I’ve heard from our customers in that space is, you know, how it’s helping some of the humans that are doing the work today. It’s a very it’s a really distressed industry, right?
If you look at commercial aviation, it’s very distressed today, extremely low profit margins, you know, rising energy prices, people having to do more with less, right? So one of the cool undertones and the cool things that I think are happening there is people aren’t necessarily getting replaced with AI. I think a lot of people have that misconception that AI is just coming for everybody’s job, right? I think what we’re seeing a lot there is people that were underwater leveraging AI to be able to actually get their job done effectively and still have a decent work life balance. Right?
James Barnes: Oh, totally. And I I think it’s it’s worth acknowledging, right, that there there is this undercurrent of massive distrust from from the workforce, from frontline workers, you know, exactly that, that people think AI is coming to take their jobs. And in some cases, it’s going to make certain tasks irrelevant, which I think we we have to accept and and discuss honestly. But the practical application of AI makes people more effective. It makes them able to focus on higher order tasks and and helps them with the stuff that I think really draws a lot of brainpower and labor and effort to maintain.
Yeah, so let’s elevate people so they can focus on, you know, higher priority tasks, higher value tasks over time.
Benjamin Tosado: Yeah, 100%. 100%. So how about manufacturing? I mean, what are what are some of the things you’re seeing there, the challenges, right? So some of the different things we’re dealing with there?
James Barnes: It’s a couple of different things. I think manufacturing is uniquely positioned to take massive benefits from automation, from AI augmentation. What we also see is this pressure where manufacturing typically doesn’t have a very strong IT budget. So if you look at AI purely from the perspective as an AI, as an IT cost center, It’s challenging for businesses to set aside, you know, funding for those initiatives because, you know, in some ways it’s viewed as an IT tool or an IT toy, right, worst case scenario. Reality is practical application of AI is not necessarily an IT function.
It’s a business outcome, right?
Benjamin Tosado: So when
James Barnes: we’re when you see manufacturing and we can do things like do predictive maintenance on your actual assembly lines, or, you know, enhance your inventory flows, or do things like digital twins, those actually produce manufacturing tangible benefits, business line, you know, business outcome benefits that really drive the company forward. They’re not necessarily IT costs. Right? So having that conversation is really challenging, but it’s very important to kind of disentangle this idea that it’s it’s purely an IT line item cost.
Benjamin Tosado: Yeah, of course. And then last but not least, distribution. I know you’ve been heavily engaged in our efforts to bring Watsco on as a customer. We’ve got a call with them today. I’m actually going to ask one of the stakeholders over there to come participate on the podcast.
Yeah. And they’re doing so much cool stuff with AI. They’re probably at at the the forefront of that space. Right? Do you wanna talk about some of the some of the stuff we’ve observed over there and some of the stuff we’re gonna help them with?
James Barnes: Yeah. And and and not I guess not to completely divulge their their, you know, their IP and where they’re going because they are super cool and super advanced. They’re doing some fascinating things and they’re true tip of the spear innovators, you know, I would see, you know, in an industry where they’re doing things like chat agents that are available via phone call, right, to provide inventory sourcing information. Right? So when you’ve got a person in the field that’s looking for a specific part, can they call and use natural language to interface with an AI model that can locate a specific inventory item or an alternate inventory item, tell them which stores it’s available, in some cases how far away that store is, what the price is.
It really can accelerate somebody in the field that would have to spend fifteen, twenty minutes searching through different portals to find something. Much easier to pick up a phone, just dial a number and get an answer. Really, really cool stuff happening over
Benjamin Tosado: Yeah, yeah. And I think, as you said, they’re the tip of the spear, I think probably a blueprint for what AI and distribution can look like. And we’re seeing a lot of cool things kind of across that space as well. So we’ve got a lot of cool things coming up here on Cleared for Takeoff. Our very next episode is going to be really neat.
We’ve got the president of one of our companies in the aviation space coming on here. We’re going to be talking about a variety of different topics, right? Both AI applied to the industries that we’ve discussed earlier. Another really cool thing that we’re starting to use in our business and helping customers build these solutions is AI assisted development. So we’ve got AI building AI, right?
James Barnes: Oh, yeah.
Benjamin Tosado: Why don’t you talk talk a little bit about that? Because I think that’s a really cool aspect as well.
James Barnes: Yeah, I think it’s fascinating. And it’s something that I’m, you know, personally using and personally playing with and I don’t come from a development background. So I think as soon as I say that, you know, some coder out there is going be terrified of what security risks I’m introducing to the enterprise, which, know, we’re taking appropriate steps. But I think what’s really interesting and unique about AI assisted coding or AI driven coding is AI knows AI. Right.
So, you know, talking about developing AI solutions,
Benjamin Tosado: using
James Barnes: AI assisted code really helps accelerate that process. It reduces, you know, the time that a person would have to just spend understanding frameworks. And obviously, like high level governance really stays at the human level. But the nuts and bolts of coding AI solutions really can be accelerated with AI development as well. Yeah, really, really cool stuff happening there.
Benjamin Tosado: Yeah. Super cool. Super cool. And that’s a great kind of segue. So I really appreciate everybody listening to this very first episode of Cleared for Takeoff.
We’re excited about the conversations that are coming ahead. I’m actually sitting down next with Jon Baker, the president of Airvoyant. We are they’re actually at the tip of the spear too. We’re working with some of the most advanced AI, Agentic AI technologies there and helping them build a first ever Agentic AI aviation parts procurement platform. So he’ll be on the phone.
We’re gonna be talking to him about that as well as the AI foundry we’re building for them and the AI assisted development methodologies that we’re rolling out there. So very excited to have everybody come listen to that. In the meantime, you can find us at clearedfortakeoff.fm or any of the major podcast platforms. Thanks for listening to Cleared For Takeoff. Hit follow whenever you’re listening so you never miss an episode with change makers in aerospace, defense, and aviation.
And to continue the conversation, find me on LinkedIn, Ben Tosado.
Key Takeaways
- Procurement was underserved, not unsolvable. Maintenance tools and mechanic-facing systems drew the attention and the budget. Buying parts was assumed to sort itself out, and the inefficiency compounded for twenty years.
- Fragmentation is the obstacle, not the technology. A vendor network that runs from global OEMs down to small shops holding the one part a fleet still needs is the reason aviation never got its Amazon effect.
- The pivot cost almost nothing already built. The workflow automation and ERP integration came first and became the foundation the agents run on. The agentic layer sits on top of work that was required either way.
- Confidence scoring is what makes automation safe to trust. The sourcing agent returns a recommendation with an honest confidence level. Humans stay in the loop, their overrides feed the training data, and automation earns its way to a threshold rather than being imposed.
- Building AI with AI changes team shape, not judgment. A foundry runs on small, experienced teams and AI-assisted development. People still decide what to build, verify what gets built, and keep governance at the human level.


