Conversational AI: Hype, Limits & Human-Centred Design (With Peter Swimm)-Ep #197

Like the best chatbot experience I ever had was I was in Atlanta, you know, one of my least favorite airports in the United States, and my flight was canceled. And I learned it from the text message that included a button to rebook three possible flights. And I think that is the best example of chatbots where it's just like, hey, it's me,

automation, you can go wait in this 20 minute line or call someone, or you can press this button that I'll book you now and you go have a beer, you know, like, right. We're going through something absolutely historic. Technologies across the board are growing exponentially.

It's a disruption that's going to completely redefine the way businesses compete. The next decade we're going to lose 40 percent of these Fortune 500 companies. The exponential growth of computing is continuing. AI is nowhere near its full potential.

Whether you like it or not, that the future cannot be stopped by anyone. Welcome back to FutureTech and Foresight, the podcast where we explore what's next for work, society, and the human experience in a world shaped by intelligent technology. In this episode, Mark is joined by Peter Swim, a leading expert in conversational AI and

the founder of Toilville, to break down where automation and chatbots are actually delivering value and where they're falling short. Peter shares his early experiences with chatbot frameworks, including his time leading product at BotKit and Microsoft, and explains how AI went from scripted FAQ bots to today's

multimodal agents. But don't let the hype fool you. Peter emphasizes the critical limitations of current AI tools in customer service, particularly when it comes to privacy, accountability and actual usefulness.

Together, Mark and Peter explore what makes a chatbot actually good, how AI agents should be designed to support, not replace humans, and why context, personalisation, and flexibility matter more than ever. The conversation also dives into real world examples from banking, healthcare, and enterprise

AI rollouts, as well as the nuanced ethical questions emerging around AI's growing role in business and society. If you've ever wondered why so many chatbots frustrate you or how we can do better, this one's for you.

About today's guest, Peter Swim is a veteran of the conversational AI world with over 25 years of experience translating complex technologies into usable tools. As the founder of Toilville, he helps organizations implement AI in a way that preserves institutional knowledge and respects human expertise.

Peter's background spans from organizing global music festivals to leading product strategy at BotKit and Microsoft's co-pilot studio. Through his work in startups and enterprise environments, he's helped shape the modern landscape of AI agents, chatbot frameworks, and human-centered automation.

His mission, build AI that works for people, not the other way around. Hi, Peter. Thank you very much for joining the podcast today. I'm very much looking forward to talking about one of the things that keeps coming up on the podcast because everybody likes to talk about it, AI, agents, and kind of the future of where we're going.

So very much looking forward to to jump it into it. Thanks for joining today. Thanks for having me, Mark. Yeah.

So the way that I like to start off all the conversations with my guests is to get a bit of a more personal sense of how you got interested in artificial intelligence and the things that we're going to talk about. was there like a certain moment or was it just kind of like a natural organic progression into doing the work that you're doing? It's kind of one of the things that was an organic progression, but looking back there

was a couple of pivotal moments. And I think like it probably goes back to like I began my career in tech in early stage startups. And so like employee number four kind of roles and people in tech know that like

at startups you wear multiple hats, you know, so you do one job but you really, you got to cover for 18 jobs. And so one of my roles was covering our support and communications and stuff like that. And I was getting at the peak of our popularity, our product was covered in Mossberg and New York Times did the thing on our product. So we were exploding popularity, you know, pretty impressive for 2006.

And so I was getting hundreds of emails a day, right? And I was just drowning in it. And that's why I started looking into like automation for these tools.

And the automation was simple then, you know, it wasn't driven by AI, it was just kind of like, here's all the boiler plates and I as a human, I'm steering the ship and I go in

and I change the beginning or change the end. And so I'm doing 800 emails a day, right? All by myself. And I could still do all my other work

because it wasn't my main job. And that was kind of a regulatory experience for me, like how technology can amplify the job you're trying to do when the job you're trying to do is complicated by the jobs

you're not interested in doing, or they're not the main part of your role. And so a couple of years later, I had the opportunity to start working

for a chatbot framework company called BotKit. And BotKit launched with the Slack API. And Slack API was a very important thing for Slack because they had like this usage,

but you couldn't really do work with it. You know, it's a place to hang out, but it wasn't until the API came along that you were able to like truly integrate it

into your work processes and have it do things. And chatbots were like the method to do that, right? Because they didn't have, it didn't have UI or buttons or anything,

but you were able to like connect the chatbot and say words to it and it would like reboot the server, right? Right.

And through that role, I have kind of like, You know, before I worked as a product owner in one vertical and one industry at a time. And from that point on, I was in every vertical in every industry, you know, doing my products and and eventually every modality, right?