Joe Green (TechForge): One of the themes that’s been cropping up on the show floor today quite a lot is obviously agentic AI, AI in general. But as my grey hair and overly wise demeanour will inform you, I’ve been around the block a bit and I’ve heard of automation before, and we used to call it, back in the day, robotic process automation, which was if you like at its most basic level, I would’ve defined it as automated copy and paste. You would copy from Excel and paste into, I don’t know, a different Excel document and then that would be the process that’d be automated. That would be RPA. So my rather controversial opener to you, Ron, is do you think that agentic AI automation is the kind of new look RPA or is it a whole different ballgame?

Ron Pierson: That’s a great question. I’ll give you a piece of little tidbit of information. So coming from Blue Prism, our founding leadership team, our CEO as part of Blue Prism, coined the term RPA for robotic process automation. So to your question though, I think they go hand in hand. I think agentic AI is the key buzzword right now and everybody’s learning, but I think RPA is still part of the story. I think it’s part of the execution alongside agentic AI. So I want to say it’s different. I think they’re alongside the journey.

Joe Green (TechForge): Okay. I mean, is it a case of sometimes the CEO has got a bee in his bonnet and he wants AI and he must have AI. And actually you guys come in and say, “Well, actually, you don’t really need AI. What you could use here instead, which would be infinitely more deterministic and predictable. It would be something else”?

Ron Pierson: It’s a conversation we have every day: CIO mandates the use of AI, it’s easy, magic wand AI, it’s going to fix a problem. But we have these conversations every day. We want to use AI, but people are still learning around AI, but it also could be expensive, and you’re talking about tokens and things of that nature. So we have those conversations. Is it an AI use case? Is it an RPA use case? But what we’re finding is a mix of both. You can use RPA for a piece of the automation workflow and AI is the other piece. Or there’s strictly just deterministic workloads you can just solve with RPA and avoid those complexities and AI, or [the] costs associated with it.

Joe Green (TechForge): Okay. And what about the platforms beginning to appear which are ‘low-code citizen automation platforms’, in that we’re almost on the brink, although some vendors would claim we’ve achieved this status already, but I would say we’re almost on the brink of being able to automate systems just with a voice command. Now, this gives us the whole concept of the citizen automator.

Ron Pierson: Of course.

Joe Green (TechForge): Do you think this is dangerous territory for ‘Bob, the accountcy assistant’ to be playing in? I mean, no disrespect to Bob’s anywhere.

Ron Pierson: Certainly. Yes, that’s a good question. I don’t know, I think we’re all learning. I think it depends what industry and vertical you’re in, whether you want some safeguards around that. But there’s been freemium models outside of AI for years, whether it be file collaboration or whatever it is. But I think there’s always a sense of consumers wanting to play around with it, but it also brings to new life and new energy and new innovation. But there’s definitely going to be some companies, organisations or industries where we’re going to lock that down. So I think it’s depending on where you’re at and what vertical you’re serving. And I think that ‘kind of depends’ [is] where my answer is.

Joe Green (TechForge): And it’s a pretty young technology, AI. I mean, all right, in dog years, it’s fairly long in the tooth, I guess. But if we ignore the kind of academic research background and imagine that AI was born in 2022 when OpenAI came into the public consciousness, we’re only talking four years. Obviously I’m not trying to talk down your business, but is there an argument that this isn’t a mature enough technology yet, really, to deploy inn anger in serious businesses?

Ron Pierson: I don’t know. I mean, I’m not a historian, but I think machine learning and AI has been around 30, 50 years. I just think now even at this great, amazing event today, everybody’s still learning. How do you scale AI? What are the best practices to put within workflows? What’s the best way to manage and corral it and scale? So I think people are still learning models and how they all fit with one another. So I think it’s still early. Well, I’ll talk from both sides of my mouth, but [it’s] still early, but there’s some organisations getting great succeses from it. So I think we’re just at the early stages, what’s going to be really the next boom as far as innovation, but it’s very fast, that’s for sure.

Joe Green (TechForge): Are there particular types of organisations where you think this type of agentic workflow works really well? I think immediately of customer services and chatbots being used instead of answering phones and directing customers the right way. That was a fairly quick win. Are there more of those quick wins waiting to happen? Are there any particular sectors that you think, “Well, yeah, that’d be a good place to deploy.”

Ron Pierson: Quick wins, low hanging fruit? I don’t know, from what I see, finance teams, accounting, invoicing, contracting, all those workflows are low hanging fruit or wins, but those aren’t as easy as what … It’s not something you just pop in an AI agent per se, and it’s up and running, it’s [not] a magic wand. There’s still some things to figure out. How is it wired in and how is it really going to bring value to the businesses versus just a thing and it works? And then from a cross vertical, what you mentioned, it goes back again to this event. You see people from small, medium organisations to organisations like HPE. So it goes across the board. And I also think it’s industry agnostic.

Joe Green (TechForge): Obviously you’re installing systems in a huge range of industries and a huge range of sizes of organisation. Are there any particular obvious pitfalls which you see and maybe roll your eyes slightly because you see them so often? Are there any particularly common mistakes that people are making when they start jumping on this kind of ‘AI train’, as it were?

Ron Pierson: Well, I think I mentioned it before, I think the common pitfall or misconsumption is like it’s the AI magic wand. It’s easy, you press the easy button, [the] ‘AI easy button’ and it’s going to magically go to work and bring all this value and ROI to the business without understanding there could be consequences, right? You want to protect yourself. You want to be able to connect that AI agent or AI into different models or different automations, and how do you orchestrate that? I think the big piece is, it’s not easy, it can’t be easy. You’ve got to protect yourselves and you’ve got to have a good framework around it, maybe the right people in place, and just really trying to understand what’s the outcome that [you’re] looking to achieve, and work backwards from there, and then come up with a plan.

Joe Green (TechForge): Yeah. I always think that one of the mirages that AI creates is because it’s so approachable, because it talks human language, because you can use human language to talk to it, it gives the impression of being this omniscient, all-seeing, all pervasive force. And so you think, “Well, therefore it can do all these things.” Whereas actually where the detail is, is where things fall down; in the checking-the-data, [the] governance, and making sure that the databases are clean, and things like that. These are the details I think sometimes that can trip people up.

Ron Pierson: Yeah, potentially I think so. I mean, that’s not really my side of the world. It’s just more about how do we maximise the ROI that I’m looking for from an AI and what’s my path to get there? And I think that folks maybe don’t have the greatest plan or understand there’s going to be some time and effort to make sure it’s successful rather than just plopping in an AI and it’s going to all be good to go.