Podcast: Play in new window | Download
Subscribe: Apple Podcasts | Spotify | Amazon Music | Android | Youtube Music | RSS
Most conversations about taming generative AI focus on guardrails bolted on after the fact. Dougal Watt takes a different approach: generate the meaning first, as an ontology, and let everything else — agents, APIs, knowledge graphs — flow from that.
His standing-room-only workshop on AI-augmented ontology creation at the Knowledge Graph Conference drew both business executives and semantic engineers, highlighting the broad interest in the need for accurate, trustworthy AI.
We talked about:
- his work at the company he founded, Graph Research Labs and their meaning-first approach
- his very popular workshop on AI safety and ontologies and guardrails at the Knowledge Graph Conference
- the implications in AI architectures of the need to balance time, risk, and regulation
- the five main approaches that organizations are using to comply with regulations:
- guardrails
- multi-agent refinement
- governance frameworks
- sandboxing of agents
- ontology grounding
- how his Semantic Agent Harness delivers accuracy improvements better than even the best graph RAG systems
- how they build decision tracing into their framework
- the need to keep a human in the loop across the ontology-building process, as well as the crucial role of human judgement in automation workflows
- how LLMs can accelerate the task of integrating data in various organization silos
- how agentic modeling permits ongoing consistency checks on competency questions
- the need for more ontologists in AI work, but also tooling that he has created to facilitate ontology work
- how good ontology methods give organizations accuracy that is tailored to their unique knowledge
Dougal’s bio
Dougal Watt is the CEO and Co-Founder of Graph Research Labs, inventor of multiple patents and patents pending covering ontology-driven declarative generation, governed AI agents, and automated enterprise stack creation. Previously IBM Chief Technologist and a global expert in information architecture. Open Group Distinguished Architect. IBM Certified Enterprise & Information Architect. TOGAF Certified. 30+ years across four continents building and fixing systems for some of the world’s most demanding organisations. International speaker, most recently speaking about AI Safety and Guardrails at the Knowledge Graph Conference 2026 in New York.
Connect with Dougal online
Video
Here’s the video version of our conversation:
Podcast intro transcript
This is the Knowledge Graph Insights podcast, episode number 55. Generative AI gives ontologists powerful new capabilities that can help automate workflows and build and maintain ontologies. But as with any powerful technology you need wise guidance and sound methods to get the most out the tool. That’s what Dougal Watt does. He has developed an approach that leverages the power of LLMs to accelerate ontology development while at the same time guiding every step of the process with human judgement and discretion.
Interview transcript
Larry:
Hi, everyone. Welcome to episode number 55 of the Knowledge Graph Insights podcast. I am really delighted today to welcome to the show Dougal Watt. Dougal is the CEO and founder of Graph Research Labs in New Zealand. He’s also the former chief technologist for IBM in New Zealand before he started his company. So welcome, Dougal. Tell the folks a little bit more about what you’re up to these days.
Dougal:
Thanks, Larry, and thanks for having me on. Great to be here. Well, what I’m doing these days, I started Graph Research Labs to explore a really exciting intersection between information architecture, which has been part of my practice for a long time, information architecture, graph, and AI, and in particular using ontologies and graph to solve some of the really major problems that I’ve seen throughout my career. Those sort of problems they … What we’ve seen over time is that most enterprise IT, it just becomes really focused on the past. You see layer upon layer of systems, often over decades. Each gets laid onto the last. And then organizations, they just end up spending more time, more effort to maintain that past than they are to build the future. And unfortunately, AI is making that worse, not better. So again, organizations are layering agents onto that same siloed set of foundations, and then they wonder why data’s not accurate or governance just doesn’t keep up and governance evaporates.
Larry:
Yeah-
Dougal:
So what we’re doing, we flip that around at GRL. We focus on generating the meaning first. So instead of the meaning being an afterthought, we start from meaning and then we move into the data and then we generate that software stack. So the idea is that you shift to modeling your business once as an ontology, and then our declarative engine generates that whole stack for you. So you can generate the things you need, governed agents, APIs, apps, knowledge graphs, MCP servers, data products. You can generate all that in minutes. So that’s been the focus.
Larry:
Very cool. And a lot of that … There’s so much in there. One, I love that it’s messy and AI is not making it any better, but everybody I hear in Silicon Valley says, “No, we can just fix that.” But that was one of the points of your talk at … Oh, quick background. We met at the Knowledge Graph Conference last month where you did this great presentation on … Was it AI augmented or LLM augmented ontology creation? I forget the exact title, but a really interesting presentation with a really well-developed workflow around how to use the capabilities of LLMs productively, but with really tight guardrails, human interventions, and things like that. Can you talk a little bit about … Well, the other thing about that workshop is it was packed. I somehow got there early enough to get a seat, which I was grateful for. What motivated you to put that workshop together and why do you think it was so packed?
Dougal:
What motivated me is I saw this extreme focus on AI and then people needing to understand … Well, I guess I’ll take it back a bit. I think what was interesting to me is that it was so packed, and there was such a strong lineup of speakers at the conference in general, but I was really genuinely surprised how much interest there was in that workshop. And what we covered, we covered a lot of ground, but it’s fundamentally about AI safety and ontologies and guardrails. So that was the focus. And I’ve never actually had a session before where you have to open up an overflow room and then people are sitting on the floor. It’s kind of crazy. For a technical workshop, that’s just really rare.
Dougal:
And I think this is one of the really interesting points is that a large number of people that were there, it might’ve been around 50-50, were senior executives, vice presidents, and their senior technical staff. And to me, that’s a really strong signal that what’s important to people is this AI safety and ontology issue and it’s being driven from the top of the organization. And that’s exactly what you want to see.
Dougal:
And I think as to why there’s that focus on the ontologies, well, everyone’s tried AI and they’ve tried different approaches, and they’re finding that getting that into production is a real challenge because accuracy’s not high enough. So if you’re aiming for greater than 90% accuracy, whatever the use case is, you need an approach like ontology. So ontology seems to be the solution and people are looking for that solution, they’re looking for tools, but obviously they’re using AI, they’re using LLMs. How do those two come together? How do they meet in the middle? How do we get value from this amazing new technology and it’s just a technology? But how do we harness it, couple it, make it safe, and be able to drive the kind of value that we need in our organization so it can move into production and be accurate?
Larry:
Yeah, I think it’s interesting. I remember even three years ago at the Knowledge Graph Conference that the budding, the looming talks about neuro-symbolic AI, hybrid AI, the combination of the two has been there. And it’s really interesting that it took three years to get to matter-of-fact, like, yeah, this is clearly the place. And it’s really gratifying to hear that there were business people in there as well, because it’s not hard to get us excited about that.
Dougal:
We love it. Yeah.
Larry:
Yeah, executives. But one of the things that came up subsequently to that talk is one, that need for safety and guardrails and methods of taming these beasts is that there’s this kind of disconnect between the insane acceleration and pace of change in the technology and the regulatory environment that drives a lot of the requirements for these things. Can you talk a little bit about that?
Dougal:
Yeah, we do a lot of work in that regulatory space, and I think the more I think about it, I think we’re kind of at a really strange point in history. So we’re sort of caught, like you say, this conundrum. It’s between time, risk, and regulation. And so as society speeds up, the risk grows larger and then regulations are expanding exponentially. So again, the conundrum, that’s making decisions slower. So in a sense, there’s a impression, there’s that disconnect between politics and regulators and technology instead of people actually working together to make life better and to harness this technology in the right way.
Dougal:
So the way I think of it as time speeds up, risk concentrates. And you can think of lots of examples of this. Trading algorithms can move a financial position in microseconds and it once took days to build that position. Or like money laundering, that chain can now move through accounts before the first alert fires off and people know about it. So what that says is that speed without accuracy, that’s not progress, that’s actually exposure.
Dougal:
So this is where this conundrum, and I think why so many business people were in that session, this is where governed AI really, really matters. If you can price a claim and insurance against a live validated view, or for the power grid, you can look at a national grid and reroute power before there’s a blackout, these sorts of problems, healthcare, using AI to make it more affordable. So none of this requires new physics, but what it does require that we need governed information architectures that public regulators and auditors can all trust. That’s really the crucial thing.
Larry:
Yeah. I live in Europe and the EU AI Act and GDPR and other things have already created a lot of that demand. Because you’re in New Zealand, how is it in APAC region and the rest of the world, that regulatory environment?
Dougal:
I think it’s a global issue. We’re talking to a lot of customers now, and this is actually the methodology that you mentioned we showed at the KGC, that’s really built for this sort of moment. We’re seeing across all sectors an increase in regulation. And look, it’s not to disparage regulation. I think it’s certainly needed in many contexts. And each rule by itself, that’s a rational decision to make. But I think when you layer them together, that’s the worst of both worlds. So the institutions start moving more slowly and they carry more hidden risk because I think fundamentally they can’t see through that sediment to where their real exposure lies, which is what this methodology was built for. And I think where ontologies and our community actually provides a really substantial advantage to dealing with that conundrum. So it’s growing everywhere. And the APAC region, definitely very similar. Lots of industries are subject to global rules anyway like GDPR. So it’s pretty much universal, I think.
Larry:
Yeah. And like all these kinds of things, they fall to IT. It’s like, “Hey, we got to be compliant. How are we going to do this?” And you’ve talked about, there’s four or five, I think five things you’ve talked about that are sort of emerging, I don’t know, best practices or approaches to managing this. Can you talk a little bit about that?
Dougal:
Yeah, sure. And this is such a new area and it’s evolving all the time, but if I think about what’s happening out there at the moment, there’s five main approaches that organizations are using to try and address that conundrum, that risk, time and compliance question. So I’ll just quickly go through them.
Dougal:
So the first one’s a guardrail layer, and it’s kind of like a boundary. So you build a perimeter around your AI. There’s lots of players here like Llama Guard and Bedrock Guardrails and AWS. Those are fundamentally, you could almost imagine them as being a sort of a filter, but things can get through that kind of filter.
Dougal:
The next one, and there’s been a lot of work done in this space, is they call it multi-agent refinement. And that’s actually something that I talk about in my methodology. This is quite an important one. So the golden rule is you don’t trust the AI or you verify everything that it does. So this refinement thing is really saying, “Let’s have agents work together, propose something, critique it, and then verify it.” And studies in computer science show you can get really good return from doing that.
Dougal:
The next layer is a governance framework, and this is the regulatory. We’ve already just talked about a few GDPRs, one example in privacy, but the EU’s got the AI Act. NIST have the AI RMF Framework. These things, they compliment each other. They require you to do things like traceability and provenance. And so they’re good standards and guidelines that we have to work within.
Dougal:
The next layer up, which is a more of a technical thing, is sandboxing of an agent. And that’s a really important technique as well. It’s saying, let’s actually technically put systems in place so that there’s a blast radius that the AI can’t go out of. So really you can’t have it calling internal systems it’s not allowed to touch, for example. So there’s a lot of work going into that space.
Dougal:
But none of those deal with, I think, the fundamental problem, which is the source of truth and accuracy in the AI system. And so this last one is loosely called ontology grounding. And the idea there is that we actually change what the AI is able to reason over itself. And so in our case, we take this into a whole new level, but the idea is that you are constraining the AI, so it can’t really hallucinate, and you are providing it information in a verifiable controlled manner. So those are the five major approaches that we’re seeing that people are trying out there in the world.
Larry:
Yeah, I love it. I’d love to drill into the last one a little bit more because this is the Knowledge Graph Insights podcast. But it’s all relevant because all of us now are working in hybrid architectures of one kind or another, and we all use these tools in our daily work. But a lot of ontologists listen to this podcast. So I’d love to dig in. Maybe talk a little bit about your approach … In the workshop, you talked about, I can’t tell you the number of machine learning engineers I’ve met who goes, “Oh, we’ll just automatically generate the ontology.” And your workshop was the whole thing of addressing that mindset. But can you talk a little bit about the benefits of LLMs and how they help, but the kind of guidance they need and more to the point, how can accelerate and amplify ontology practice?
Dougal:
If you like, I’ll share the slide actually from KGC because I think-
Larry:
Oh yeah, actually, most people just listen to this.
Dougal:
Okay.
Larry:
But send me the slide. I’ll share it in the notes so we can talk about it. Because I’ve got your deck too, and I can just grab that if needed.
Dougal:
Oh, okay. Yeah, cool.
Larry:
Yeah.
Dougal:
Yeah. So that neuro-symbolic approach or the ontology grounding approach, it’s really interesting. We built a thing, and I mentioned this in the workshop called the Semantic Agent Harness. And I like an analogy for this because it’s a very technically complex area. But in a sense, what we’ve done is that we’ve put a bridge into the LLM and it’s got a door and a lock on it. And that in fact, several bridges. And those bridges really define what the AI is allowed to go out to, what it’s allowed to access, and we can gate and control that. And we also put almost like a transponder on the AI because working in this regulatory world, you have to make sure that you can prove something did what it said it did. And so the transponder records this whole provenance audit lineage sort of chain, and you can go right into the AI’s prompt, how it reasoned the reasoning trace, the data and the data out … You can get all of that data out and record that for regulatory purposes.
Dougal:
And then the other side of the bridge is this whole set of tools that we have, and they give it a lot of capabilities to the AI. They give it things like memory. They give it data that’s grounded in the ontology. And that’s really the crucial core part of it is that the ontology is the source of truth and the ontology’s grounded and linked into enterprise data so that if you need to fan out and retrieve something, there’s a channel for doing that. And like I said, that bridge and that door and that lock, that’s all gating and controlling what the AI has been allowed to access and how it goes out and does it. And we’ve done experiments with this where if you use the most advanced forms of RAG, GraphRAG, you can get into the 90%, but we can get into the very, very high 90s and in most cases 100% accuracy using our test framework. We have a whole automation suite to automate and test these competency questions and how they apply to the AI and how it reasons over its data and retrieves data and responds.
Dougal:
And so you can get to a extremely high accuracy level. In fact, probably actually more accurate than if your AI was answering some question on your internal data and your regulatory posture, here’s the law and here’s our internal framework, how we respond to that. And the AI is passing that data and responding back to workers in the company or people outside the company, it’s going to probably be more accurate ultimately than a human because it’s been grounded so thoroughly and controlled so thoroughly. So that’s really, from the work we are doing where we see that whole area going, and it subsumes a whole lot of the previous approaches, I think, that it’s just a superior approach to doing it fundamentally.
Larry:
Yeah, it’s such a learning process and it’s great to see things coming together like that. Hey, I want to revisit one thing. When you talked about the transponder, it’s almost gathering this provenance and lineage data. That evokes for me that thing that I was introduced to by the famous Foundation Capital article about decision traces.
Dougal:
Yes.
Larry:
Is that an element or a component of the system you’ve built? And I’m inferring that it is from what you said about the governance compliance, but …
Dougal:
Yeah, absolutely. I mean, it’s so cool because we’re effectively injecting into the operating context of the AI, of the LLM. So these agent frameworks now have a lot of hooks that you can hook into. So we can get back the whole decision trace. So we gave it this, it did a set of reasoning activities in response to a prompt, and then it returned some data. And so that whole context and that whole reasoning trace is actually we pull it out and we can store that and we can actually use it as memory for the AI for self-improvement as well. So yeah, it’s a lot of really, really interesting stuff you can do there.
Larry:
That’s interesting. I wonder too, I haven’t heard too many people talk about this, but I wonder if you’ve done anything like the other kinds of decisions you can trace. I’ve done a lot of enterprise UX and service design work where it’s just so clear that there’s so much valuable information just being lost in Slack channels and emails and meeting notes and things like that. And so much of what you showed was extracting entities and information from those kinds of corpuses. That’s another … It seems like the built-in transponder-based decision tracing, that seems almost like a no-brainer at this point. But I know the harder work is always capturing that tacit human knowledge and doing something with that. Do you try to tackle that as well?
Dougal:
Yeah, and that can happen. So Slack’s one example. Another might be that we have a set of use cases, like a banking one is a good one. So a typical use case would be we bring into the equation, we’ve modeled out what we call an accelerator pack, and it’s a set of pre-built ontologies that define a legal area. So a bank has to respond to a set of laws about, for example, anti-money laundering and know your customer and a whole range of other Basel II, Basel III. And then the bank produces a compliance framework that responds to the legislation. And so that comes from their internal control framework. And we ingest that through our document-integrated tool and we link it ontologically. So now you’ve got this really rich knowledge about here’s the laws, here’s how we respond to them.
Dougal:
And so people can be working inside that organization and they might submit a query to the AI and they might come up with something, “Oh, look, I’m building a new product. It’s for SME lending that’s in this area.” And then the AI can say, “Well, here’s the controls that fit. These things are all good. You need to comply with this.” And then so I have this dialogue in this use case, and I might surface something through my dialogue that’s new, that’s genuinely novel or that’s missing. And so that signal from the enterprise, kind of like your Slack example, that signal to say, “Well, here’s an area where we don’t have coverage and we need to ground.” The AI knows, and it’s forced by our harness to not make things up. So one of the gates and controls is that if it queries the knowledge base and it gets nothing or gets a null result, it can’t say, “Well, this thing exists.” It says, “This is null. We don’t have something that helps you in this area.”
Dougal:
And that signal that goes back to a human, and it gets back to the workshop where one of those key aspects of the methodology is that a human has to be in the loop the whole time. You can’t validate and trust unless you’re in that loop. And so in this case, that worker has surfaced something really useful, a signal that’s not in our ontology or our instance data right now. Now we’re going to use that signal where a human’s going to look at it and go, “Yeah, there’s a gap there. We genuinely need to address that.” And then you move forward and the organization evolves as a whole, the machine and the human sides of it.
Larry:
I love that. It seems like a level above human in the loop. It’s like, “No, we got humans all the way along,” which has always been … I had a guest on way back early in the series, with Yaakov Belch, an ontologist in Israel who says, “Humans in the loop? No, humans in control.” And you’re enabling that, it sounds like, with your system. Now, one thing about that, in this atmosphere where there’s this insistence at high levels that everybody’s going to be using AI, and one logical outcome of that is that, “Well, just replace people.” And so how do you sell architectures that require human intervention across the workflow to executives who are maybe not on board with people doing a lot of stuff?
Dougal:
Well, look, I think … I heard that described this way, which I think is great. There’s two sides to a business. There’s operational. And what you’re talking about there is a cost takeout. “Oh, we can just make people redundant and take cost out,” which is not the right way of thinking of it. The right way to think of that side of it is how can we make these people more efficient and get more done in the day? And when you start doing that, this is a very human thing and it’s an economic law almost that if you give people more of a technology or a service and you make it cheaper and more affordable, which AI is doing, we will consume more of it.
Dougal:
So all of a sudden on a given day, those workers can get more done in that operational side. And a lot of the focus in AI is there right now. And there’s a lot of useful things that you can do with AI to automate workflows and processing and speed it up and just make it generally more efficient. But people need to understand that AI and agents, they’re not self-directed. A human has to ultimately direct them. And the latest craze is this thing around loops. “I can get an AI to do a loop.” Well, we’ve had that for over a year, and some of us have been building with it, just do that as a natural practice anyway. But again, you have to define outcomes, you have to define what good looks like, product management practices that we’ve had forever. And so you’re not making people superfluous, you’re making them actually a critical part of your automation story because ultimately a human’s got to say, “Yep, that makes sense, that doesn’t make sense.” Or, “Wow, I didn’t think of that. Let’s move forward with it.”
Dougal:
So I just see it as another way of generating value and just speeding things up. And also if it’s harnessed in the right way, the other way to think of it is that AI is one of the most perfect language translators or language understanders that we have. So much of our interaction is unstructured data and passing decisions and discussion and talking. And now we can actually bring some level of automation into those workflows. And with the right type of harness, we can actually make it really, really accurate, extremely accurate. And there’s just so much value to be had from that, I think.
Larry:
Yeah, that notion. Just this morning on LinkedIn, I read one of a dozen posts I’ve read in the last two days about the fact that automation has been around forever and that there are other ways to do it besides gen AI. And I guess do you have heuristics or approaches to evaluating what the best way to automate a process is?
Dougal:
Every use case is kind of different, but if there’s a whole sequence of steps and they’re relatively data-heavy and they’re not yet automated, that’s a good heuristic for something that an LLM can contribute to improving the cadence, the speed of that workflow. We do a lot in that regulatory space, for example, and in healthcare, and those are great examples where there’s a lot of complex data flowing around. Well, in fact, actually … I mean, I sort of said this at the start, this is organizations. They have all of these data silos and all of this tech layered on top of these data silos. And to this very day, integration, simple use cases like integrating across those silos is very complex and very hard to do.
Dougal:
And so AI, if you think of a heuristic, being able to automate, or at least if I’m pulling data out of all of these silos and being able to understand the landscape alone, there’s huge value in that. So anytime it comes to complex data processing, complex data workflows, unstructured data coming in, those are all really great candidates. But again, it has to be grounded in ontology because the LLM will hallucinate from Chinese whispers from one end of that silo to the other, all those silos you’re trying to aggregate up that there will be drift over time in the AI. And so you trust but verify or don’t trust and verify whichever way you look at it. And that’s what that ontology grounding and the agent harness, that’s what that’s all designed for.
Larry:
Yeah. Something you just said that reminded me of something I wanted to revisit from earlier when you talked about there’s something in your workflow where you revisit the competency questions. And it occurred to me that how often do competency questions just get put up on the shelf at the start of a project? And how often do people actually come back to them? And this kind of gets at, I don’t know, the notion of observability and just systems improvement. Was I inferring correctly that you … What do you do with that when you revisit the competency questions? Are you constantly fine-tuning them or suggesting new ones? Yeah.
Dougal:
So that’s a great one about … It’s a knowledge-heavy, complex thing to create them. And you’re right, most people, they might have a go at writing them in the old days and they just sit on a shelf and they never get revisited. And the ontology goes through so many iterations. And this one’s up here and this is shot way over there, and so they’re completely out of sync. And this is the beauty of agentic modeling, the workflow that we talked about in the workshop, that you probably have a set of core invariant competency questions. Things like, if you have to comply with the regulation or law, it’s pretty invariant what you have to do for some of that stuff. But then how you respond to it and the internal processes and the workflow that you have, those sort of things, there’s a lot of institutional knowledge in an organization about how they deal with those sorts of things.
Dougal:
So your competency questions get split into some invariance and some things that have probably higher cadence of variation. And then some competency questions which are purely things like analytics and reporting-focused and maybe delivery-focused. And so as you go through that GRL methodology that we talked in the workshop, you have to re-sync and you have to be quite smart about how you re-sync against those competency questions. So you don’t want your ontology … It’s a dance. You don’t want that ontology to drift away from the core invariant ones. One day the ontology, the LLM goes and deletes a legal obligation class. That would be a terrible outcome. You don’t want it to do that. So you have to pin it back and say, “Well look, I’m just going to rerun these.” Maybe they’re in SPARQL and I just get Claude code to execute them for me against the ontology. And it goes, “You just deleted something that this thing hangs off.” And so yeah, really important, I think.
Larry:
Yeah, no, that’s awesome. I love it. It’s sort of, I’m just increasingly thinking about how this new gadget in our toolbox is advancing our practice as well. So it’s gratifying and cool. Yeah.
Dougal:
Yeah. I mean, I think there needs to be more ontologists, there needs to be more ontologies and there needs to be more work done in this space. And if we can use these tools, like I say, to take a piece of workflow that actually used to be hard to do, mechanically hard, and just … Who’s got time to synchronize things and go back and spend five days revisiting some CQs? If I can get the machine to just do that check for me as I go through my workflow, and it’s just part of the harness that this whole thing’s being driven in, that’s fantastic.
Dougal:
But at the end of it, again, it comes down to a human thing. And one of the tools that was in the workshop that we showed is the OntoLinter. And that’s a really, really critical piece of software. And the version that we put up for KGC, we are opening that up so people can just use that if they like. That’s a crucial thing because you have to be able to retest at lots of different levels when you’re an ontology engineer using an agent. So if I flow my ontology through the process, every time I jump back to another cycle, every time I exit one of the steps, I should link to my ontology to make sure, just like code, it’s saying “I’ve done a sanity check. The structural and the semantic things aren’t broken.” It actually makes sense as an ontology. That’s a really, really crucial thing.
Larry:
Yeah. No, I love this. It’s a little bit like eating your own dog food, but it’s sort of like the good version of, if all you have is a hammer, everything looks like a nail. If all you have is an ontological approach to things, you will figure out an ontological solution to your ontology. Anyhow, I love that. Hey, Dougal, I think-
Dougal:
Well, actually-
Larry:
Yeah, go ahead.
Dougal:
No, that’s just one other point. That’s a really, really important thing because I think that historically for us as ontologists, that community to test their ontology in the real world, that was a really expensive thing to do because it requires highly specialist skills. You have to know SPARQL, all those sorts of questions. And we built our generator tool to generate things like APIs directly off an ontology because that practice of, okay, you’re right. I’ve done a great ontology and I really love my ontology. Can I test it in the real world with real data that’s integrated with real systems? That’s one of the most crucial questions. Does it actually meet a set of use cases and business value? And the only way you’re ever going to know is if you can put it into some …
Dougal:
Flow it through into that production sequence. It’s not in production yet, it’s in pre-prod and I’m testing it with real data and that’s why we built that generator. I can create an API and I can flow data in, nobody needs to know SPARQL, the standard teams working in the company that build their standard software in their standard ways with REST APIs and what have you, microservices, they can participate and they can help me to validate and verify that my ontology meets the needs of the organization. That to me is a really, really crucial step.
Larry:
Yeah, no, like I said, this is an amazing time to be in this work. But hey Dougal, I can’t believe it. We’re coming up on time.
Dougal:
Sorry.
Larry:
It always goes way too fast. Always happens. But before we wrap up though, I want to make sure that you get a chance, is there anything you’d like to revisit from the conversation or just want to make sure you share before we wrap up?
Dougal:
I just think to my mind, that methodology that we use, I think that’s really key for using ontologies in the age of AI. You’re going to need that accuracy. And from all the work that we’ve done, this is really ontologies and this type of methodology, it’s really the only and the best way actually to get that accuracy that’s really customized for the organization. And you can’t trust AIs, that’s why you need this critical thinking. You’ve got to keep testing, you’re going to need tools to do that. I mentioned OntoLinter is a great one that people can access. Other people are making great tools too, but tooling and automation is just critical here. As we spin that flywheel faster and faster and we get more done, tooling and automation everywhere that we can put it in to keep our ontologies accurate and grounded and keep our AIs where we want them to be.
Larry:
That spinning flywheel reminds me of what we were saying earlier about, I’m pretty sure this is not going to replace people. It’s just going to create more cool stuff for them to do. Yeah. Hey, one very last thing, Dougal. If folks would like to connect with you or follow you, what’s the best way to find you online?
Dougal:
Yeah, sure. So people can just connect to me on LinkedIn personally and just connect there anytime. Or you can go to our website, graphresearchlabs.com, and there’s a contact page there and there’s some resources that people can look at. And yeah, look, people, please reach out. Love to hear your use cases and what people are doing.
Larry:
Excellent. Well, thanks so much, Dougal. I really enjoyed the conversation. Great to see you again.
Dougal:
Great to see you. Thanks.
