The Deep View: Conversations

Jason Hiner (00:01.27)
In this episode, I talk with Dave Horton, Airia vice president of solutions engineering. Dave and the team at Airia have been working diligently to make AI safer in the enterprise. is a platform that lets organizations plug AI into all their systems and still keep it under control so that AI can make an impact without leaking secrets or going rogue. Dave and I also talked about the huge surge in interest in AI agents in 2026.

and what that has meant in the enterprise. Here, Dave has a bit of a contrarian take on open claw and the wild expectations that some companies have for agents, especially when it comes to replacing human labor. We talked about how managing AI in the enterprise comes down to a combination of three things, security, governance, and orchestration, and how interoperability with other vendors and solutions is one of the bets Area is making to stand out.

Dave and I also talked about his journey through three companies with the founders of Aria, why funding has exploded for AI startups, and how the AI ecosystem is unfolding in London, the UK, and Europe. Dave also shared the AI tool that's dramatically changed the way he works. So here it is, our conversation with Dave Horton of Aria.

Jason Hiner (00:01.934)
So Dave, for those who aren't familiar, tell us what the problem that Aria is trying to solve in the world and in the AI space and what's your role at the company?

Dave Horton (Airia) (00:12.755)
Yeah, absolutely. think what area does is evolving with the markets. so initially we started out solving the problem of as a business, how do I build AI quickly? How do I prototype? How do I see if it's going to be a good use case for me to take into a production scenario? And so we started with this product called orchestration where it's how do I build effectively? How do we integrate with my data?

What we've seen though is that that's maybe the first hurdle for many organizations. The next hurdle is well, can I secure it? So, know, when it comes to AI security, there are new threat factors. And so, you know, we stood up a security AI product as well that would help enterprises not only build, but also secure that AI. And then more recently, we obviously very close with customers, especially my team.

But we are also looking at the governance of AI. Now it's not enough just to say we build with AI and we now secure it, but you also have to be very defensible about, are we using it for the right reason? Do we meet local and global frameworks for acceptable AI use like the EU AI Act and such?

That's really what we do is we're anywhere there is an AI problem. We're building product to solve that for customers. And that's evolved over the last two years or so. My team and kind of leaning into what I do with the company, what my team does, we are essentially the customer facing technical team for the area platform. And so my team probably has the highest touch point with customers when it comes to implementation of AI. And so, you know,

Equally, it's exciting because we can kind of be part of the solution that customers are looking to implement. And so we can work with their kind of business units and look at use cases that are really going to transform their business. But equally, there's a lot of education mismatch in the market. And so we're also there to kind of set some expectations around what is and isn't possible and what it will take in terms of investment to get something that will be truly useful when it comes to AI.

Jason Hiner (02:31.576)
Very good, so this space is getting a little bit crowded too, right? A lot of companies are now saying like, they will help you to keep agents safe in the enterprise. Since in the last 90 days, agents have had this sort of amazing burst onto the scene with personal AI giants and open claw and perplexity computer, all of these things and more to come. And so,

there's a little bit of some sort of rogue concern about rogue activity with agents in the enterprise, which brings, I think your company's like area into the purview of a lot more people. And at the same time, there's a lot of companies running around saying, this is dangerous, let us help you do it. how do you talk about what you all do in context of everyone else that's also sort of claiming that they'll help you get agents under control?

Dave Horton (Airia) (03:25.551)
It's kind of interesting, I guess, you know, any market where there is a lot of innovation and a lot of concern and, you know, a lot of kind of continued education required in order to kind of keep up is actually really ideal market to be in, right? If you can sell the solution to multiple problems, that's even more ideal. And so...

You know, as a company trying to keep up to date with the latest and greatest, not just from what everyone else is doing, but also, well, how do we secure some of these innovations that have come out from Anthropic? And, you know, how do we respond to like a mirror fish and an open claw kind of set up in an enterprise? You know, it's given us some good...

collateral to go after and kind of reiterate, hey, these things are happening, whether you like it or not, you know, either embrace a solution or, you know, be fully prepared to either be stifling innovation or, you know, not protecting yourself adequately.

Jason Hiner (04:31.842)
What does kind of the open claw moment mean has meant for you all since this has sort of raised the awareness of agents, but also just maybe AI security in general because it's gotten a lot of publicity around the risks.

Dave Horton (Airia) (04:48.271)
think OpenClaw and Mirrorfish and whatever the next thing is going to be, I think they all have a foundation in hype. So, you know, there's this kind of, you know, one YouTube video that spreads into 10 that spreads into thousands and, you know, the actual practical use cases, I think, arguably are questionable, you know, in a...

in any kind of person setting, like is that really useful to have a dedicated system that is running off and spending tokens on your behalf? But I think what it does do is it ignites a bit of a kind of a FOMO moment for many enterprises. if we should understand this, if we don't, are we gonna be left behind by our competition that might be leveraging this to do some kind of, you know, clever,

kind of angle with this new technology. And so Mirrorfish was a very similar one where it was kind of selling on the predictivity of particular scenarios using LLM. And it's like, well, that sounds like magic. You know, if we don't do, we don't explore, if we don't experiment, then, you know, what does that mean? You know, we could really, we could really innovate here. But the hype versus the reality is often, you know, the, it's too late to realize whether it's useful or not. You've already met.

made some waves in terms of potential damage to enterprise data, enterprise systems. And so we often will lead with, hey, let's help you innovate carefully and safely. Like we'll provide guardrails that would, you know, through our gateway, you could essentially set up a open claw and actually analyze what is actually doing, what is actually spending.

and also put some constraints around what it can and can't do. And so you can put it in a bit of a sandbox that allows you a bit more control. But I wouldn't say that we're professing that this is a great idea for enterprise and it's something you should definitely do. I think there's a part of the role that we have is that we're really here as a value partner to our customers to sort of help them with what they want to experiment with.

Jason Hiner (06:53.166)
Mmm.

Dave Horton (Airia) (07:06.412)
But ultimately just do that in a safe way, give them the tools to do that without risking their own intellectual property, their customers' data, etc.

Jason Hiner (07:15.214)
Very good. you all are controlling AI agents and orchestrating them the way that you describe, know, basically making sure they can meet your compliance standards, your sort of corporate governance standards, laws, regulations, all of that is a big part of it. Is your platform primarily focused on agents or a little more broadly on AI in general and helping enterprises get it under control?

Dave Horton (Airia) (07:44.427)
Yeah, I mean, initially we were building a platform that would rival many of the other orchestration platforms out there. so, one thing that we've learned in the last 15 years of building innovative companies is that actually customers don't just want one thing, they'll eventually want 10 things. And they're not just gonna be using your product, they're gonna be using other people's products for some aspect of it. And so when we built the security platform,

Jason Hiner (08:04.674)
Hmm.

Dave Horton (Airia) (08:14.438)
it was actually with the purview of let's not just secure the area side of things, but what else also if we could secure the other orchestration platforms. So if we build a gateway that you can put guardrails and then you can put other orchestration platforms pointing at our gateway, they can benefit from our security capabilities as well. And so this has really lent us into a market where we don't really care what the customer's using.

If there is a possibility, we'll help them technically govern and secure those investments, regardless of whether they were built in an area in the first place or not.

Jason Hiner (08:53.356)
Interesting, so you've gone the way of more interoperability and sort of pragmatic approach to this versus the sort of walled garden approach that you have to sort of use our thing and if you don't want to use it or don't want to use parts of it or you want to use somebody else's thing, it doesn't work as well, which is a little more, I think the world has been moving a little more toward that more apple walled garden moment. So what is that like for you all? Why did you make the decision to?

to sort of focus more toward interoperability with other tools.

Dave Horton (Airia) (09:25.193)
think working with enterprises, if you look at some of our customers, they represent some of the most scalable companies in the world, some of the most secure. And so one company that might be perfectly fine with a cloud setup and using public foundation models, for example, are actually probably far and few between. It's more likely that I want a wall garden and I want to have...

Jason Hiner (09:46.254)
Mm.

Dave Horton (Airia) (09:49.852)
control over my data, I have a security standard that rivals some of the biggest, most secure companies in the world. And so these sorts of conversations are not alien to us. Like we're fully prepared to have a conversation with a customer that has to be completely walled garden. What we do offer though is the flexibility and it's more than kind of a guiding.

principle that we'd say, well, you could go completely wall garden. You'll miss out on maybe some of these interesting use cases where you use a finite foundation model, you leverage MCP or something that is maybe more on the new end of the integration spectrum, for example. And it's really up to them. We provide them with the menu. It's almost up to them to tell us what they want. And so we're

kind of fortunate that we can be on premise. We can be private cloud, we could be dedicated or shared cloud. so equally all of the components in our platform, the customer can own the data or they can host it with us or they can host it in third party. We give them the kind of prescription and they tell us exactly what they would like to see in a final product from us.

Jason Hiner (11:08.782)
Yeah, very good. do you have customers, examples of customers that either they can name or can't name, that are really using your platform to accelerate what they're doing in AI? Because we know that there was the survey, not survey, but study that went around last year that 90 % of companies, very famously, are not getting.

ROI out of their their AI deployments and projects. It was a little skewed, right? think those results are a little bit overstated, but we do know that there are companies, there are lots of companies that are struggling to kind of move from proof of concept to deployed AI solutions that are driving a lot of value. And do you have examples of companies that are using your platform to really help them bridge the gap?

Dave Horton (Airia) (12:00.815)
Yeah, I mean, it's interesting when we get engaged with a customer, you think it would be very, very well prescribed, like how we would go in and kind of pitch the products. But ultimately every company is at a different level of maturity with AI. Some of them are only just scratching the surface right now and all they've really seen is from a consumer perspective.

Others have been developing for 10, 15 years, maybe more from a machine learning angle and now they're kind of leveraging generative. Part of the challenge is that everyone's at a very different state of preparedness to have that conversation. But also AI is not a technology problem. It's actually a business problem is typically what I see. When we go in and we speak about

Jason Hiner (12:53.518)
Mm.

Dave Horton (Airia) (12:55.192)
you know, let's integrate AI into your business. You can't really spend all the time talking about the area product because it's actually relatively simple to build an agent and LLM, connect it with tools and data sources. The bit is, is like, well, where did you get the data from? Like who is that person that goes and gets that data, where they put it and where do you want the output from this? What do you want that to look like? And so...

you when you look at a typical engagement of a use case, it might be actually a very small slice is actually the area kind of AI component. And the larger proportion is well, you know, the business transformation, like the business kind of end to end process that we have to kind of embed in and kind of make that as seamless as possible. So when it comes to, you know, why did POCs fail? think, you know, on the face of it, it sounds quite negative to AI when

you hear these sorts of statistics. But I'm actually quite encouraged when POCs do fail because it means that they're taking risks. They're saying, well, I think it could do something like this very well. And it might not fail because the technology failed. It might fail because it was too risky or it didn't give us the value that we were expecting. So.

Jason Hiner (14:02.83)
Hmm.

Dave Horton (Airia) (14:17.538)
It's actually a positive sign because even if you launch 100 use cases and 80 % of them fail, but you succeeded with 20 of them, that's something that has an impact potentially massively compared to the other 100. I think as well when POCs fail, might be down to actually AI was not the tool to solve this problem.

actually sitting down in a room with people and understanding what that end-to-end process was. so, like I say, it's not a technology issue when they fail. It's more likely a business or a people problem.

Jason Hiner (14:59.436)
Yeah, boy, it always comes back to that. doesn't. It always comes back. Well, so area has been around or come out of stealth, I guess, for about 18 months ago. You said that you have over 300 customers in the enterprise, enterprise customers. You recently raised $100 million in funding last fall, fall of 2025. And so

Dave Horton (Airia) (15:02.367)
Hahaha

Jason Hiner (15:27.01)
What is the founding thesis been for Aria and how has that gone? What is the current kind of trajectory and how has all of the things that Aria set out to create and do come to pass?

Dave Horton (Airia) (15:45.28)
It's interesting, I actually came from our Q2 company All Hands directly before this call and it was kind of highlighting the last quarter, which was the biggest quarter ever for us. We're actually at 500 customers as of last quarter, so we've extended. And it's not just small companies anymore, it's big kind of household names. They're looking to us for assistance in their AI adoption, which is fantastic. But I think...

Jason Hiner (16:01.518)
Okay.

Dave Horton (Airia) (16:14.621)
You know, the founding thesis, you know, this is the third company that the leadership team area have founded and grown. The first two, you know, very similar. They were really designed to solve particular problem for our customers in enterprise. So the first one, AirWatch, was really focused around mobile security. Right around the time, you know, the iPhone came out and

people started getting their corporate email on iPhone. Before that was Blackberry. the problem we were solving was, how do I get my enterprise email onto my personal device? And how do I secure that adequately? And all of the risks that go into that. That was obviously a big kind of a-ha moment when it comes to, we've got a solution that can handle this, not just for iOS, but any device.

Jason Hiner (16:59.054)
Yeah.

Dave Horton (Airia) (17:13.022)
The next company was one trust which was squarely in the privacy and compliance space. so, you know, having the DNA of a security platform, very, you know, closely linked with consumer trend as well. Having that background of compliance and understanding of data sovereignty issues, global frameworks, it's kind of lent itself very well to what we're building here at Aria.

product that is very much being innovated from a consumer standpoint, being integrated into enterprise and very closely linked with security and governance. so I think we've probably got an unfair advantage to maybe some of our competition that haven't got the same pedigree. But I think the founding ethos is we build a blueprint where

The platform that we build is not just suitable for just legal in mid-market in Germany, for example, we're not painting ourselves into a niche. It can literally be adopted by any customer, any size in any country for any use case. And so it's lent itself very well to being something that can be discussed with any CISO or CIO, like where they're looking to innovate and secure.

Jason Hiner (18:36.878)
Very good, so areas sort of combining this sort of three things as I understand it, know, orchestration, governance, and security. I there's a lot of overlap between those. You know, a lot of times companies are kind of bolting those things on afterwards after they've started building. How do you manage that when you're working with a client and you're also getting the opportunity to help.

clients with sort of some of the transformation process. Like, okay, you know, we can help you with the AI part, but in order to do that, you have to build some better foundations. That's sort of what I hear when you talk about some of the things you're talking about.

Dave Horton (Airia) (19:14.597)
Yeah, I mean, you if you consider we started orchestration, we built security and then governance. you know, a lot of the time it's because of customer feedback. It's like, we're, you know, can I secure my AI? What about shadow AI? Like, I don't, there's a lot of AI that I don't know about my business. How can I govern that? And so, you know, we are very close to customers when it comes to, you know, what are they hearing? What's the pulse check that they have?

Jason Hiner (19:24.556)
Okay.

Jason Hiner (19:32.621)
Yeah.

Dave Horton (Airia) (19:43.618)
around AI in their business. so to a degree, like our early customers and arguably even the ones we sign in the last quarter or so, we're not just signing them for revenue benefit, but also they can kind of steer us, they can tell us what we should develop. And so if there's something the product can't do, then we're in a really enviable position in the market because about 60 % of the company is dedicated to R &D.

Jason Hiner (20:04.844)
Hmm.

Dave Horton (Airia) (20:13.815)
we can just feed that back into the engine and say, well, you know, that's not impossible to deliver. We may have already, you know, got some of the foundational parts in place that we can leverage. so, you know, customers are really on a journey with us when it comes to AI. And for them, it's a benefit that, you know, they can speak to us and, you know, if it's a really compelling use case, if it's really something we can see scaling beyond just one or two customers, like something that

hundreds of customers could leverage, we can put the energy into getting something out the door very quickly from a product standpoint. They get the benefit of having something in the product that they're using that can solve another problem that they don't have to buy another point solution for. it's, again, the unfair advantage we have is the pedigree we have, but also the amount of team that we have dedicated to the build process.

Jason Hiner (21:13.592)
How big is the team?

Dave Horton (Airia) (21:15.544)
So the company's running just shy of 200 today. And if you look at each area, we're not just tied to a particular geography. We do have teams in, I believe, 11 countries and offices in several of those. it's not just like North America we're kind of focused on. We've got a big presence in Europe, the Middle East, where we've seen a lot of funding for AI.

you know, again, that region's really been pushing us in a way that means that the rest of the world can kind of play catch up and we can deliver products that are really key for those use cases. it's 200 people, you know, having this follow the sun sort of model means that we get insights into regional requirements, data sovereignty issues that might be very specific to a particular region. It makes us a much more

a mature platform that we can implement for a large enterprise, which also operate in multiple geographies potentially.

Jason Hiner (22:21.848)
Yeah, okay. So you mentioned, you know, this pedigree and sort of experience that the team has going back, you know, multiple products. Are there a lot of folks, you know, in among the 200 and area now that have been part of like all three companies?

Dave Horton (Airia) (22:37.354)
Yeah, it's actually quite interesting. you know, myself, I was part of AirWatch and OneTrust before Aria and obviously I've been with other companies since, but, you know, I certainly, one of the people that has kind of seen what we've built before from like a, you know, a seed funded company to a, know, series A, B, C, D, whatever. And so that is the case for quite lots of the employees, either from...

Jason Hiner (22:43.49)
Okay.

Dave Horton (Airia) (23:05.973)
you know, the AirWatch perspective or from the Wantrous side, you know, there has been, you know, a kind of kinship that you can bring to people, especially if you've worked with them in the past. Working in a startup is obviously very different to working in, you know, Fortune 100, but, you know, I think there's a special type of person that you need to hire to make that successful. But equally, you know, when...

to get to 200 people in 18 months and have such a highly effective product, it helps if you hire people that you've worked with in the past and know the drill. They've seen the blueprint work. You don't need to educate them into what they're joining and what they sign themselves up for. It's gonna be busy. It's gonna be fast paced. And so, that is, I guess, a part of the blueprint for what we've built.

Jason Hiner (24:00.974)
Very good, how about, what were you doing before you started sort of with this group of folks, these startups, how has your career evolved to this point?

Dave Horton (Airia) (24:14.483)
Yeah, I I was obviously a young man 15 years ago when this was all kicking off. know, when I left university, I moved into kind of healthcare and government. you know, it was interesting. It was at that point where, you know, I got a taste of, well, you know, what are the products that I need to secure those agencies, like from a government and healthcare setting?

Jason Hiner (24:18.158)
Hahaha

Dave Horton (Airia) (24:44.375)
And that's how I got introduced to AirWatch before ultimately going and working there. so, since then, there's always a natural kind of flow in and out of startups. So as they grow, they're no longer startups, they become large enterprises. And so you start to gravitate towards other companies that might pique your interest in other technical areas. so...

I've worked in the Salesforce ecosystem more recently before joining Aereo. I worked for a technology reseller, selling security portfolio products where I again, kind of leverage some of the DNA from the AirWatch and the One Trust. yeah, I think I've been quite fortunate obviously to be on this journey, but it's...

you know what you're signing yourself up for and you know it's going to be exciting when you get a hundred million dollar kind of round of funding and that's not even a series A, there's obviously a big commitment to the future of the company and to how we can move very quickly.

Jason Hiner (25:57.401)
Love to get your perspective on that. know, it's funny, we see all these companies in the AI space coming out and they're, know, some of these rounds are huge. Like we're talking numbers that just are pretty recent that we've seen sort of hundred million dollar, you know, funding rounds coming out of stealth. What's your perspective on that? Why are we seeing, is it the scale of the opportunity? Is it the enthusiasm? Is it the fact that it's very expensive to run companies?

Dave Horton (Airia) (26:06.802)
Mmm.

Jason Hiner (26:26.914)
doing this, all of the above, what do you think?

Dave Horton (Airia) (26:30.316)
I mean, VC funding is very much focused on AI as it's clearly the dot com of the era. think if there's anywhere to put some money, if you get it right, you can get it very right. So if you get it wrong, then I guess that's the risk you take. But what other industries could you say are this transformative?

Jason Hiner (26:41.262)
Yeah.

Dave Horton (Airia) (27:00.224)
lot of cash that was someone else's. Where are you putting your money in? Spreading it across AI companies is probably a pretty safe bet given the trajectory of what we've seen in the last few years. But also, coming out of the COVID era, VCs weren't necessarily putting money into software and they can't sit on it. It's not growing at all. So they've got to spend it on something.

Jason Hiner (27:03.758)
You

Dave Horton (Airia) (27:28.749)
You know, it's certainly the market that has maybe captured the most amount of digital transformation potential. And yeah, if you invest correctly, then, you know, the sky's the limit on what you could get as a return. So I guess, you know, from that standpoint, I think that's really why we're seeing such a lot of revenue kind of pushed into companies such as Aria that's...

you know, the potential is really there that could see it be the biggest companies in the next 10 years are going to be AI companies, I'm sure.

Jason Hiner (28:02.446)
Yeah. So, Dave, you're based in the UK, near London. What's the AI ecosystem look like in London, in the tech ecosystem right now? As you and I talked about, I've spent some time there as well, and it's had its ups and downs. The scene, I remember the Silicon Roundabout era where there was just kind of a lot of things happening. yeah, with the AI space, of course, Google DeepMind is based in London. That has some cache to it, but...

What do you see in the UK and in London and maybe even in Europe broadly in terms of the AI ecosystem?

Dave Horton (Airia) (28:39.629)
Yeah, I you know, I think if you look towards OpenAI and Anthropic, they've made some significant investments in the UK. You know, they can't just exist in the US and, you know, arguably as well, if you look more towards Europe, companies like Mistral and Cohere, like they're very much the LLM for the non-US kind of arena as well. there's always an angle.

And there's always a benefit that AI companies will have by extending a footprint into Europe. More so because they can, the closer you can sell to those regions, the better you'll do. Especially if you've got a presence in Germany or in the UK, then it's gonna be easier to sell to customers in the UK or in Germany. So the other angle as well is,

Fundamentally, there are differences in business setup when you look at, certainly Europe is more highly regulated than the US market, for example, and sometimes the choices that the enterprise will make is actually more based on the values of the data protection, the data sovereignty that sit behind the vendor.

Part of the reason why we've invested in the UK and into Europe as well is that we can now sell very effectively to those regions. We've got language skills. We've got also part of that pedigree that I mentioned. We've always run the companies, a global company. It means that we can speak the local language. We can speak the issues that are really facing those companies. so again, it makes it a much easier sell cycle.

given that background.

Jason Hiner (30:36.558)
So one of the things that you and I talked about a little bit too is that when we talk about AI right now, especially in the enterprise, really and in consumers as well, it gets reduced to just LLMs. But what you all have seen is that...

the AI ecosystem has a lot more diversity, a lot more nuance than just LLMs. I'd love to hear a little bit of perspective, like what areas learning on that and what that looks like and how it's sort of shaping what you all are doing.

Dave Horton (Airia) (31:11.114)
Yeah, I think the difference between an LLM and an agentic business is quite a gap. It's not fully understood by customers when they start speaking to us. so one of the aspects is that LLMs have no context of your business context. And so the foundation model doesn't know anything about your org structure, your policies, your

data taxonomy, the compliance obligations that you have, for example. And so, the first step is typically, well, how can we make AI speak your business, your business language? And so, having that grounded in your own enterprise specific knowledge through retrieval of mental generation, fine tuning and orchestration, this really governs how the adoption will go if they're not leveraging this.

but then, you know, the, I mentioned like the consumer wave is often ahead of the enterprise wave. So, you know, roll back to when GPT first came out, for example, you know, it was, it was a novelty. I wouldn't necessarily say it was, it was prime for enterprise, but it was certainly, you know, an interesting, technology that people adopted firstly as a consumer. Very similar to, know, when I mentioned about.

enterprise mobility and like the iPhone coming out, like all of this was led by consumers that seeing a very interesting kind of technology that they want, they wanted to leverage. And then it's like, okay, enterprise comes along and they're like, right, how can we do that consumer thing in the enterprise setting? And so that's really where the security and the governance kind of play a bigger role. Whenever you do something as a consumer,

It might not be at the forefront of security and governance in your mind when it comes to that innovation, but certainly for enterprise, it's much more of a forethought than an afterthought or you'd hope.

Jason Hiner (33:18.542)
So circling back a little bit to agents. agents of course are having this major hype moment as we talked about earlier. When it comes to enterprise...

what are enterprises actually ready for and sort of deploying in practical ways versus what's still mostly aspirational and hype? I'm sure that area has a perspective on what's working with agents and what's getting deployed and turned into valuable kind of ROI versus what's the stuff that's still a little bit pie in the sky.

Dave Horton (Airia) (33:56.966)
Yeah, mean, whenever I speak to customers, there's a little bit of, know, it's kind of the problem is we walk in, it's like, what do you want to do with AI? And it's like, you get some really unusual kind of science fiction and, know, it's like, right, okay. So you can almost anticipate where, you know, where you're going to have a challenge almost before you've started implementing. But I think the...

Jason Hiner (34:07.8)
Yeah.

Jason Hiner (34:11.254)
Yeah

Jason Hiner (34:21.806)
Mmm.

Dave Horton (Airia) (34:25.189)
The context that we're trying to convey is it's okay to have big aspirations when it comes to AI. It might be a very comprehensive system that you need to build to get that full end to end. A lot of the time it's gonna be iterative. Like version one is always gonna be the worst version of your agent. It certainly is for me. I wanna get that version one to kind of a quick POV.

And then I'll add on some tools. I'll maybe add an agent that is dedicated to just looking at, know, calendar invites on my calendar, for example. I'll build another agent that is dedicated to document generation if I need to build a report off the back of it. And so it's kind of stepping back to, you know, class-based development where you're building basically pockets of code or systems that are dedicated to one particular task that you can reuse again and again.

So when we talk about agentic AI, a lot of them, a lot of customers getting in at ground floor and thinking, wow, this is this, there's so much to do. But when you think about it is you're just defining individual pockets of what you would like to build as a, as a system. And you're building agents around that. And then ultimately in the future, you don't need to redesign your agent every time to give you a nicely formatted document that you're expecting for like a board report.

I've already got those agents built. I just need to connect them in the right way. so, you know, orchestration is really around not just how do I build an AI agent with an LLM and tools, but how do I get agents speaking to other agents where they've got specialties and they have their own skillset that have already been defined, already been ratified by the business and already secured with the governance as well.

Jason Hiner (36:16.142)
What's the biggest gap between what companies think they're getting from agents or will get and what they actually get?

Dave Horton (Airia) (36:26.198)
Yeah, think I'm still calibrating myself, I think in some regards, but I think there's misconceptions that, I can replace an entire department of people if we build an agent. And it's like, you know, I can certainly see there are areas where you could make efficiencies, but I don't think that that should be the reason to push forward with an AI initiative.

if I look at my role and how my role has evolved, you know, two years ago, I was not in AI. you know, I was, you know, probably more from the old school in terms of the tool set that I would leverage to do my day to day. and if I look at how my working life has changed over the last two years, it's really with the purview of, okay, I'm using AI as a co-pilot. You know, I'm doing the work, but I'm instructing AI to take the mundane.

out of the role that I'm doing to a degree. And so I think I'd rather businesses start to look at, well, how can we get smart people working on smart problems and remove some of the monotony around those tasks they don't like to do? you know, ultimately it's keeping people happy, but it's also making sure that you're using AI in a way that is still human controlled, that there's still a human on or in the loop.

for the decisions that come out the back of it.

Jason Hiner (37:54.584)
Very good. How about what's the one AI tool you're using right now that's sort of changing, you know, the game for you, making you more efficient or, you know, helping you in whatever way.

Dave Horton (Airia) (38:03.488)
I think one thing that's been very recent for me, I've not traditionally been a developer, but I can code. I've been always technical in some regard, but the ability to use Claude code to build a working prototype on the full stack deployment, that's just something that would have been science fiction five years ago. And now...

everyone's using Claude code to kind of create, you know, working applications. And so I think the, terms of a concept, that is the biggest transformation I've seen in, you know, the way that I work is that I don't just explain to someone else how I want something to work. I can literally vibe code it myself. I can test it through. And then when I'm explaining to someone what I think we should build, I can provide them a working prototype and the concepts right there, you know, it's,

very different way of operating, especially in the software industry, which, yeah, like I say, that has been the biggest transformation, probably the most sci-fi element of using AI in the workplace for me.

Jason Hiner (39:17.07)
Very cool, well Dave, thank you so much for your time. Great to be here talking about this stuff with you. It's so top of mind for so many businesses, so many professionals right now, so appreciate it.

Dave Horton (Airia) (39:31.167)
Absolutely, it was a pleasure.