Most companies want AI for the same reason: it makes shareholders happy. Chris Huntingford thinks that's the wrong starting question. In this episode of The Team Check-In, host Emre Ok talks with Huntingford, Director of AI at ANS and a Microsoft MVP, about what gets lost when organizations chase AI for its own sake, starting with a fact that stopped Emre cold: a single prompt in a tool like Microsoft Designer produces roughly the same emissions as charging a phone.
Chris breaks down the real difference between predictive, incremental, extensible, and agentic AI, and makes the case that most companies build AI for what the technology can do rather than for what the person on the other end actually needs. He also explains why he now treats AI as an idea starter rather than a productivity tool, how EU regulation is already shaping what agents are allowed to do with people's data, and why the real fix for irresponsible AI use is education, not fear.
.svg)
.svg)
Emre [00:00:07] You're listening to the Team Check-In, powered by Teamflect. Hi, everyone, welcome to another episode of the Team Check-In. And with me today — I have an incredibly fun voice when it comes to digital innovation. Chris Huntingford — is it Huntingford, two names, or Huntingford, one name? But yeah, man, I'm not precious about it. Honestly, no one ever gets it right, it's crazy — I live in England and no one gets my surname right. I'm like, guys, it's Huntingford, Huntingford.
Emre [00:00:39] Okay, yeah, but it doesn't matter — honestly, I'm used to "Huntington," "Hunting-field," "Huntingford," fabulous, whatever you like, loving it. And Chris, you were in the UK very recently — you were in Scotland?
Chris Huntingford [00:00:53] Yeah — but you told me you're not an avid whisky drinker. So what were you doing there,
Chris Huntingfo [00:00:58] dude? So I was at a tech conference called Scottish Summit. It's run by two guys, Mark Christie and Ian Connolly [names as heard; spelling not verified], who started it five or six years ago, and it survived through Covid. It began as a few hundred of us just showing up, and now it's close to a thousand people. Two of my friends, Zoe and Matt, have taken it over now. But yeah, a bunch of nerds getting together — then we went to explore Aberdeen after geeking out. And, man, I just suck at whisky, purely because — I've tried it a few times, and I think I spent a lot of time as a kid stealing my dad's whisky, which ruined the taste of it for me. So I'm willing to learn, but yeah, I'm not good at whisky.
Emre [00:01:39] Yeah, I mean, I want to appreciate it really badly. Whenever people can find the notes, smell it, taste it, whatever — I'm always really impressed by that. But what I sometimes don't get is when someone takes it, sniffs it, takes a sip, and goes, "Hmm — oak, vanilla, and burnt tire." Burnt tire? Is that good? How?
Chris Huntingford [00:02:04] Like, how did they decide that? It's crazy, though — I'll tell you a hilarious story. I was nearly thrown out of a bar in South Africa by one of my colleagues for this. We'd won an award, and he said, "Cool, man, I'm going to buy you a shot." I said, "Awesome" — I come from a bit of a sketchy background, so I'm thinking, oh, a shot of tequila, cool, let's go. And he buys me this glass — I'd never seen this before, probably worth about £20, never experienced it — a glass with ice and whisky in it. So I go to smell it and think, oh man, this smells like a barbecue — or what we'd call a "braai" in South Africa. I'm like, why does this smell burnt? And he says, "No, it's peated — it's supposed to smell like that." I said, "Okay," and then I asked the bartender for a Coke and poured it into the glass and started drinking it. And he's like, "What are you doing, man? You can't do that, that's sacrilege — you've just wasted a 500-rand shot." Rand, don't get me started — but that's genuinely expensive for South Africa.
Emre [00:03:04] Then why are you guys taking it as a shot if it's —
Chris Huntingford [00:03:07] He said, "No, you're supposed to sip it." I said, "Okay, but then why —" So yeah, he was mad. Anyway, I drank it, didn't actually like it at all, so I ended up pouring it into a pot plant. So, to anyone listening — I'm not endorsing that, honestly, it wasn't a good thing to do. But you know what, I just can't drink
Emre [00:03:27] whisky. All right — so, other than being a person who feeds whisky to succulents, you're also a Microsoft MVP. And I love this line from your bio: "Microsoft creates the dots, and I connect them." Can you elaborate on that?
Chris Huntingford [00:03:50] Yeah, I can. I'm actually an ex-Microsoft person, and I always say — how do you know when somebody works for Microsoft? Because we'll tell you. It's a bit weird, right. I think I've graduated from being a support-and-maintenance person — someone who'd literally go to customers and fix problems on Microsoft technology — through enterprise architecture, and now focusing a lot on AI. I took a step back at one point and thought: what does Microsoft actually do? They're not very good at implementing products. They're extremely good at making cool things, and they know how to sell them, but they don't really implement them. Have you ever seen Microsoft implement Office? You implement Office. So I like the concept of dot collection and dot connection. Some time back, one of my mentors — I believe her name was Donna Osaka [spelling not verified] — said to me, "You're very good at connecting dots after you've collected a bunch of them." And I thought, actually, that's a good point. I'm very good at collecting a lot of information and getting people to share what I need to know. Then what I focus on is connecting those pieces of information into a full story. That's a genuinely fun thing to do, in my opinion, because I like puzzle-solving, and I think other architects out there would agree. So yeah, that's what I do — it's the easiest way for me to explain it.
Emre [00:05:15] Awesome. And how did the MVP thing come about?
Chris Huntingford [00:05:18] This is actually a bit of a weird story. Microsoft has this program — there are about 3,500 MVPs, Most Valuable Professionals, in the world. It recognizes people who are super active in the Microsoft communities, doing technical evangelism, running events, getting people on board. I didn't actually know what I was signing up for when I started. I thought, "Cool, MVP" — and I did plan on getting the award at first, but then I couldn't, because I just wasn't good enough, or whatever. So I stopped planning for it and just got more involved in the community, and really enjoyed it. I recently did a session at Scottish Summit about how the Microsoft community impacted my life — I genuinely believe I wouldn't be where I am without those people. But it kind of happened by accident: I went for the award, stopped trying for it, and got more involved in running events. One of my friends, Will Dorrington, and I spun up something called "Those Dynamics Guys," which was really cool. Microsoft started getting involved in what we were doing, we kept running events and making connections, and then, out of nowhere, I got an email saying I'd been recognized as a Most Valuable Professional. I was like, "Cool — do I get a sword? Flames? Is there a unicorn? What happens?" What's actually amazing about it is that the program is designed to mentor people into leadership-type roles, where you get to work closely with some of the really senior people at Microsoft, which is honestly a real privilege. And also to help your friends get into the program — I'd bring people on board and get them using Microsoft tech, and it's really fun, because most people stick around in the community and make amazing friends. Honestly, the most rewarding part has been building this huge group of friends I see all the time, go to events with — it's like a little tech family. So that's what the award really is. But like I said, I originally planned to get it, and the moment I stopped planning for it, I got it. I don't know what to make of that.
Emre [00:07:38] More often than not, when you actively participate in the community and are genuinely helpful to people in the Microsoft ecosystem, people recognize it. One thing I think we're both really passionate about, especially in the Microsoft world, is implementing a digital ecosystem inside a company. We're deeply in the Microsoft ecosystem here at Teamflect, especially through Microsoft Teams and Outlook, and we help companies digitize their performance management and everyday processes. So, in your opinion, what's the secret to building a successful digital ecosystem for a company?
Chris Huntingford [00:08:28] Oh man, I love that question. I had this conversation once with a friend of mine called Bora — I can't remember his surname for the life of me, but he did something pretty amazing. Let me try to spell it — I think it's S-E-N-K-A-Y-A [spelling not verified]. He came up to me after a session once —
Emre [00:08:58] Did you just name-drop a Turkish person to another Turkish person? Yeah, that's Şenkaya, probably.
Chris Huntingford [00:09:04] Yeah, that's a Turkish name, right. So — he does this in his spare time, he studies neurology, that's just what he does for fun. He's a Power Platform maker, dev, or architect by trade, but studies neurology on the side. He came up to me after the session and asked about ecosystems: "What is the definition of the ecosystem around you — your ecosystem?" And I thought, that's such a good question, because I think the ecosystem you build is the direct area you have control over. This gets a little "sciency" — we talked about it for a long time, by the way, and I'd encourage people to look this guy up, he's absolutely legendary. He said: think about it — if you're an ant walking over an apple on the ground, and you want to get some of that sweet stuff out of it, that's the ecosystem you have control over. But if you're on the Hubble Space Telescope, looking down at parts of the Earth, you don't have control over the completeness of the Earth — just the area around you. So it's a little different for everyone. Your company's ecosystem is the digital space made up of the people, process, and technology you have direct control over. I really liked that, because it helped me localize what ecosystem enablement actually means. I tell people all the time: you might be part of a three-person company, and that's great, because that's your ecosystem — your three people, your processes, your technologies. Or you might be at a company of 100,000 people but working in a department of 200 — that's your ecosystem. It's the area, from a digital, process, and people perspective, that you feel you actually have control over. And what I've found is that setting up that ecosystem starts with recognizing that — recognizing your area of control, or "influence" is probably a better word than control — and then asking, how do I make this better? What can I improve about the people, process, and platform here? What we've found is that technology is a really wonderful catalyst for doing that. So when people ask, "How do we adopt Power Platform, or AI?" — it comes down to recognizing that you might be interested in the technology, but what processes and scenarios will it actually affect, and what people will it affect? That's roughly how I define it. Your immediate area of impact is where you want to focus — you don't have to look wider if you don't want to, that's fine, but for now, focus there. I know it's a long explanation, but it matters, because it's really about localization.
Emre [00:11:45] I think that's a great way to explain it. What a lot of people do is get lost in the weeds — there are so many amazing solutions out there, so many choices, and the second you start researching, if you're doing it on Google you're bombarded with SEO content and ads. Somehow, someway, you fight through that, land on some page, and find yourself surrounded by sleazy salespeople like myself.
Chris Huntingford [00:12:19] And, yeah — you're not a sleazy salesman who attacks you with every feature under the sun. But it's very hard to stop yourself and think: how many of these do I actually need, and which of these features will actually impact my surrounding area, my ecosystem, like you said? More often than not, what people end up with is clusters of apps — all for different purposes, all with amazing capabilities — that don't integrate well with each other, don't play well together, and now employees have to bounce back and forth: "Oh, my tasks are here, that's cool. Oh, we use this app for recognition, this one for that, this one for that." Too many things to track, the barrier to entry gets too high, and you have to spend so
Emre [00:13:12] much time on training and implementation that you're lost.
Chris Huntingford [00:13:16] Yeah, exactly. And I think this has happened with AI a bit too — everyone's going, "We need AI, we need AI now," and I go, "Cool, but why? What are you actually after?" And they say, "Oh, are we going to AI that thing? Where's the AI, where's the AI?" And I'm like, dude, it's too broad. It's the same as saying, "I'm going to do sports" — like we were talking about before we started recording, maybe there's a blooper reel in this somewhere —
Emre [00:13:49] we're definitely adding those parts in.
Chris Huntingford [00:13:52] Yeah, there'll be a blooper reel at some point. Like we were saying about sports — it's the same as someone going, "Hey, I'm going to do sports today," and you and me going, "Well, dude, you're a surfer, I'm a skater — what sport are you going to do? What have you done to prepare, have you trained?" And they say, "No, we're just going to do sports." I feel like companies right now are saying, "We're going to AI," and I'm like, okay, but in what way? What do you mean, and why?
Emre [00:14:19] More often than not, the "why" is that it gets the shareholders excited,
Chris Huntingford [00:14:25] right? Yeah, people
Emre [00:14:26] want to see it, want to hear it — if a roadmap comes out with nothing AI in it, people will freak out, the shareholders will lose their minds. So — first of all, I don't think we're a couple of Luddites here, or denying how incredible the developments are. But the problem starts with AI for the sake of AI, AI because "we have to do it, whatever it is, it has to be AI." Our website needs those little sparkly-star icons in it, fast.
Chris Huntingford [00:16:03] Yeah, exactly — the spinny sparkly star icons, you know what I mean? Sorry, I just had a flashback to using Canva earlier today, and I'm like, man, it's everywhere. I'm making a note right now to tell people to watch out for the spinny sparkly star icon, it's everywhere. But it's right, you know — what are we actually doing this for? I was at the gym this morning, and a guy says to me, "Dude, what's the best video editing product with AI?" And I said, well, most video editing products have an element of AI now — what do you actually want to do? He said he wanted to auto-trim his videos as soon as he loaded them, looking for the "ums" and "ahs," like how you'd edit a podcast. And I said, sure, you do get things like that, but my question is: is that what you really want to do? You might want to focus on getting this right manually first, and then move to an AI-driven approach. It's exactly the same as writing a document — you want to make sure you understand how to write a document first. Once you've understood that, then it's fine to bring in AI. But never understanding that, and trying to do it from scratch with AI — I can't imagine that working. That's where organizations struggle, trying to take the lazy route out, and it's not the right thing to do.
Emre [00:16:35] Yeah. And more often than not — this has been a prominent pattern on the Team Check-In for regular listeners — I come from a literature background, so with AI, I can't help but take the sci-fi outlook on it. That's all fun and good, but talking to more tech-savvy people like you, with more technical backgrounds, I'm starting to understand more of its actual implications and where it's actually going.
Chris Huntingford [00:17:14] So —
Emre [00:17:16] in your opinion, Chris, how do you see the responsible use of AI going into the future?
Chris Huntingford [00:17:24] That's a big question — there are a couple of elements to this, and I don't want to be doomsday about it, because I'm really not a doomsday person on AI. I think we've called this out many times, even in movies — what was science fiction is now science fact. That's just what it is.
Emre [00:17:41] But I —
Chris Huntingford [00:17:43] feel like humans, even with the best of intentions, can be stupid. And the reason is that we have this uncanny capability of turning walking sticks into weapons — that's just what we do. We manage to screw things up so badly, out of selfishness, greed, wanting to make money. That's not just true of AI, it's been true of automation, it's been true of the light bulb, of nuclear fusion — "Oh, we just got nuclear fusion, let's make a bomb." Come on, really? You have an infinite energy source and we're making bombs with it. So, aside from all that — we have to admit we're inherently not great at controlling our greed. So I think the future of AI really comes down to education. We have to educate people that using artificial intelligence responsibly is genuinely important, regardless of the technology. AI is an incredibly powerful tool — it's a tool. People need to understand this isn't some magical thing that just appeared, it's a tool built from other tools. If we learn to respect artificial intelligence up front, I think we have a much better chance later on of releasing it responsibly into the wider public. And I think there are forces within the tech community, the cultural community, the arts, trying to push people toward that. So that's one thing. The other thing — and I'm going to say a slightly naughty word — is that we're using AI for the dumbest shit right now. We've turned it into a pub game. "Hey, let's go into Designer and make a giant walrus attacking a city with a flamethrower." And I'm like — I
Emre [00:19:36] did nothing wrong with that.
Chris Huntingford [00:19:37] just find that funny, sure — but it's crazy, because we've turned it into a pub game, right? People are writing pirate poems in ChatGPT. But what I don't think people really understand is the actual repercussions of using these tools. As an example — did you know that writing a single prompt in Microsoft Designer produces roughly the same emissions as charging your mobile phone?
Emre [00:20:01] There you have it. There you
Chris Huntingford [00:20:03] go. So that's education, right — we don't know that every prompt we make uses the same GPU capacity, roughly, that it takes to charge your phone. Same resources. So what I started thinking is: what does that actually mean? Responsible AI isn't just about not turning walking sticks into weapons — it's actually knowing when to use AI, and not putting it into everything that doesn't need it. AI for the fun of AI isn't okay. That's the thing I'm trying to get across to people now — we really need to think about why we're doing this. As an example, with tools like Power Platform, someone once told me, "I need to know why somebody is making an app." And I said, honestly, I don't — because I don't ask you why you're making an Excel sheet, the resource utilization is what it is, you pay for that in Azure consumption or whatever. But with AI, there's a lot of stuff under the surface that you don't see. Here's a scenario: remember when Uber landed in a bunch of cities around the world and charged you, basically, mate's rates — a trip to the airport or the station might cost you a fiver? They absorbed those costs as the cost of doing business. What was actually happening is they were prepping you for a price hike — getting you used to using Uber. They did it in Seattle, they did it in the UK, they're doing it in my area now. And I think that's exactly what's happening with AI. We're being charged a micro-premium for the prompts we're using now, but once we're used to it, we'll be willing to pay whatever, because it's made our lives easier. And, again — is that responsible of the companies doing it? Not being open about the real costs and the sustainability impact? So, this —
Emre [00:21:55] I think the term for that is "enshittification." I don't remember who coined it, but I like it.
Chris Huntingford [00:22:01] Yeah, I like it too — can you spell that? I want to look it up. "Enshittification" — and the way it works is, you start providing a service, you get people dependent on it
Emre [00:22:15] by, you know, absorbing large losses upfront —
Chris Huntingford [00:22:17] yeah,
Emre [00:22:18] and over time, once everyone's hooked and all the competition is dead, you hike up the prices and lower the quality of service.
Chris Huntingford [00:22:32] This is so good — I'm looking it up right now. I think it also gets described as a kind of classification. This is so good, I like it — there's a nice pun in there with the engineering term too. And we've seen it with all sorts of services — Google's a huge example of it, same with Netflix and a bunch of others. Once you've got everyone locked in and no competition, you jack up the prices and lower the quality. But if we take this whole approach to the world of work — everyday work life, and since a lot of the people listening come from an HR background, but really the everyday workflows of people generally —
Emre [00:23:22] Yeah — what, in your opinion, are the best-case scenarios for where the use of AI goes, and what are some of the worst-case scenarios?
Chris Huntingford [00:23:32] Worst case — I still think irrelevant image generation is up there. I wish you could just block people from doing that. I understand the need for creativity and iteration, I really do — and this is where I'm torn, because you wouldn't stop an artist on their fifth iteration of a painting, you wouldn't block a songwriter on their tenth iteration of a song. So why do it in AI? But the impact of what's happening is far larger than someone crumpling up a piece of paper and throwing it in the bin. So I'm honestly still torn on it — I don't know if it's a "bad" use case exactly, but I do think there are irrelevant use cases of AI — things like pirate poems and that kind of nonsense — that I just wouldn't do. I get the creativity, I'm a creative myself, but I can't justify it, knowing what I know now.
Emre [00:24:31] Yeah, I think you're looking at it from an eco-critical perspective, and that's genuinely not part of the popular discourse around AI right now — and it really needs to be, if you think about it.
Chris Huntingford [00:24:44] It's education, man. It's real — after I found that out, my perspective changed a lot, because now I genuinely think twice. And it's not — the scary thing isn't that it's a physical, tangible thing, like putting your phone on charge, or flying in an airplane. It kind of is, and it kind of isn't. So I don't think there are necessarily "bad" use cases for AI in the sense of hurting somebody, creating harmful imagery, irresponsible use, hacking, jailbreaking — those are obviously bad uses of AI. You shouldn't be using incremental AI in tools like Copilot to go dig up your boss's salary, that's not okay. You shouldn't be using it to plagiarize a book from someone else — that's just bad form. But there's also the sustainability piece, which we have to educate people on. I think the really strong use of AI — and I've actually shifted my thinking on this, because I used to think a really good use of AI was furthering automation, and I don't think that anymore. I have a fairly significant neurodivergent span, to the point where I multi-thread — I think of it like an eight-lane highway, I'll be talking about one thing and then jump lanes, and sometimes even on podcasts I'll say something and two minutes later say something else, and people go, "Wait, what did you just say?" and I have to find my way back. But I also struggle to get started — I procrastinate, because I'm an overthinker, and I get into this huge well of overwhelm with data. So I struggle to start. I use AI now to get me going. And, dude — I don't love the word "productivity," I think it's kind of a nonsense term — but the way I work now is different. I'm far less fussed about starting something than I used to be. Do you know what that's done for my anxiety? It's wild. Let me give you an example — this morning I knew I had to do a bunch of video rips for one of my big customers, and normally I'd be freaking out, thinking I don't have time for this. But with the tools available now, I knew I had the time — I wasn't procrastinating, I knew exactly what I needed to do, and I could do it extremely quickly. So for me, as an overthinker and someone who's neurodivergent, that's taken a huge well of work-related anxiety away — and I think that's one of the smartest uses of AI out there, because it's not just changing the process of how somebody works, it's actually changing the way they live. The impact on how a person lives is far more important than just making their job slightly better.
Emre [00:27:33] Yeah — just using it as an initial spark, an initiator, that's pretty solid. And that ties in well with something we do here at Teamflect. Like you said, everybody wants to put AI into their product somewhere, and for us it came down to a real discussion: okay, we want to do this because we have these amazing tools now, and we want to take advantage of them, but we don't want to do it just to have sparkly icons on our website. We had long sessions on what we're actually struggling with, and how we could use large language models to help us and our customers in their workflow. One thing we landed on, and got great feedback on, is performance review comments. If you're conducting performance reviews and you've got a lot of direct reports and you still want to leave genuine comments for everyone, you don't want fully AI-generated comments — that's disingenuous, that's bad. So instead, when you're writing your own comment in the text box, you can enhance it — say, "This is my general idea, but lengthen it, shorten it, formalize it," or, "Hey, check this for bias — did I actually use biased language here?" We got great feedback on that. Our users also really loved our development-goal suggestions — after you've conducted a performance review with our product, since it's integrated with all your goals and feedback, our tool scans your scores, your goal completion rates, and the feedback you've received from peers, and says, "Looks like you're struggling with this, this, and this — here are some potential development goals." But again, those are just suggestions — the reviewer has full control over it, and can pick, tweak, and add whatever they want to their own comments. The idea is always to enhance the human element and make life easier for the human running things, without actually taking over.
Chris Huntingford [00:30:13] I think that's genius, dude — because let me tell you, as somebody who's done those types of reviews, I'll be sitting there going, "I genuinely don't know how to phrase this," and spending way more think-time on it than I should. I think that's a really good use of AI — to say, "This sounds good, but how would you enhance it, how would you make this better?" I can absolutely see that use case, especially coming from review-heavy organizations —
Emre [00:30:52] a very nice way to frame it — not going to say which one I'm in right now, but yeah, big global systems integrators and vendors have to run reviews that way, because you're dealing with hundreds of thousands of people. It's not the same as where I am now, which is much more personal — we have about 700 people. I came from a place with around 80,000.
Chris Huntingford [00:31:11] If you're a manager handling reviews for that many people, how do you keep context? How do you keep things personal? You can't, really — it's basically impossible past a certain number. And the other thing is context-switching between reviews — if it's you and me, we've got some similar interests, we're both into tech, we're passionate about the same stuff. But if you've got a developer who never goes out, never really talks to people, but genuinely wants to grow their career — how do you handle that scenario versus one where someone's naturally open and sharing? So yeah, I can see it. I think too many companies focus their AI on what the AI can do for the tech, not on what the AI can do for the human — and I think that's the reframe people need.
Emre [00:31:59] Yeah, that's a great way to look at it. And I think, more often than not, when we look at problems with institutions or certain practices — whether it's how we implement AI the right way or the wrong way — if you trace it back to the root cause, it's like the Scooby-Doo villain reveal: pull off the mask, and — oh no, it's corporate greed again.
Chris Huntingford [00:32:35] Oh, dude, I love
Emre [00:32:36] it. More often than not, like we said, it's "we want AI in this product because we want to keep the shareholders happy," and everybody wants to hear the word AI somewhere. I think Ed Zitron has this great concept — he calls it the "rot economy" —
Chris Huntingford [00:32:58] where the
Emre [00:32:59] rot economy is where everyone's so focused on growth — growth, growth, growth — that if you're not showing growth, you're somehow unsuccessful. This is a huge pattern in tech right now, especially — where year over year, if the spreadsheet doesn't show growth, you're considered a failure. And with just a bit of common sense, you realize: that model just isn't sustainable. It's straight up not sustainable.
Chris Huntingford [00:33:35] I love that, I'm going to keep that one — that's going to stick in my brain. I think the concept of "growth" gets defined in a lot of different ways, but naming that pattern — I really like that.
Emre [00:33:48] Yeah. And what you get at the end of the day is mass layoffs, because at the end of the year companies still want to show those numbers, and the fastest way to artificially inflate them is: "Let's lay a bunch of people off, cut costs, and look — growth again, we made more money, we lost less." And that whole mindset of constantly needing to grow, constantly needing to keep shareholders happy, is what drives companies to stuff AI into as many things as possible — because that's the "in" thing right now.
Chris Huntingford [00:34:29] But it's the wrong thing — and that's exactly why the same happened with automation. Remember, when robotic desktop flows came around, everyone said, "We can automate everything," and then people started asking, "But should you automate everything?" What you're going to see now is the same pattern with what I'd call "agentification" —
Emre [00:34:48] can you go into that a bit —
Chris Huntingford [00:34:49] AI is going to change, and just to qualify that — I don't think the
Emre [00:34:53] the concept of agentification, for people who —
Chris Huntingford [00:34:58] yeah, I'm glad you asked that, this matters. Think of a copilot versus AI generally. You've got predictive AI, and then you get different types — differential AI, where the AI changes something fundamentally. Predictive AI predicts something that will happen. Incremental AI aids you in a task you're already doing. Differential AI changes the way you work fundamentally, by doing X. Agentification sits somewhere between what I call extensible AI — taking a product and extending it with AI, not baked in, but bolted on, like building an intelligent app — and incremental AI, which is baked in, like Copilot inside Microsoft Word. A copilot is like a chatbot with no context and no memory — if you've ever seen the movie 50 First Dates, I'm going to quote my friend Donna Osaka [same name as earlier; spelling not verified] here: think of it as not knowing who you are, a really enthusiastic intern with no moral compass. It'll go, "Cool, here's an answer, here's another answer, here's a RAG pipeline that'll generate something" — it's just throwing information at you. An agent is more specific — it has context, it has memory, it understands who you are, and it has the ability to learn. An agent could, in theory, take on the role of a human. Physically, it could do that — but under European legislation, that's not going to be allowed. An agent is really about automating automation, in a generative, genuinely smart way. As an example, if you generate social media content for a living, you could have an agent generate that content for you — but under EU legislation, you still have to observe, review, and interact with the output. So I think agents are going to change the game. And we've already been prepped for it with incremental AI tools like Copilot and ChatGPT — I think these agents are a fundamental step change, and I believe that's actually what stops us from using AI for dumb stuff. I think it's going to be: here's how we supercharge our processes and scenarios with artificial intelligence. That's what I think.
Emre [00:37:31] All right — great explanation, especially with the 50 First Dates comparison and the very excited intern, I love that analogy. So, with agentification on the horizon — and not such a distant horizon —
Chris Huntingford [00:37:51] Yeah —
Emre [00:37:52] taking this all the way back to our podcast and our listeners specifically — how do you think the world of human resources and performance management will be impacted by the arrival of agentification?
Chris Huntingford [00:38:12] I think anyone in the world of administration, HR, or scheduling is going to end up with the biggest role of all in the observability layer, because you're now dealing with people. Whatever your current role is, if you're in HR and dealing with people, your role is going to shift toward making sure these agents are doing the right thing. Take onboarding as an example — onboarding somebody is easy, you can fulfill that with automation pretty quickly. But when you're dealing with something like a disciplinary process, you can't be using AI for that. You can't be profiling people with AI. And you'll notice, in the EU AI Act, something else came out recently — they were talking about a sentiment-analysis prohibition. That effectively means: if somebody responds to a message and you've got sentiment analysis wrapped around it, are you even allowed to do that? Is it like Minority Report, where you're guilty until proven innocent? What are the actual rules? So the moment you're dealing with people, AI requires real care. When you infuse AI into those layers of your business, it's extremely important that you have proper red-teaming, that you've gone through the right transparency layers, and that especially in the workforce-planning layer, this becomes massively important.
Emre [00:39:48] Yeah, especially — and the role of oversight there is undeniable. And I think — this is almost the "water is wet" of the AI conversation at this point — but whatever AI tool you're using is only ever as good as its dataset. More often than not, people see AI as some kind of unbiased observer, an unbiased way of doing things, and that's just not the right way to look at it, because there are so many occasions where it reflects overarching societal biases we might not even be aware of.
Chris Huntingford [00:40:38] But that's how this has always been. Take an example — look at Disney movies from the '50s through the '80s. There are a lot of interesting biases baked into those movies, not on purpose, it's just where society was at the time. Certain movements didn't really exist yet in the public conversation — there wasn't really visibility for the transgender movement, for instance. It might have existed, but it was never really flagged or represented up front — and now, looking back, you can see it just wasn't
Emre [00:41:07] well represented, let's say. Perfectly said. And now, when you look at the norms of society, there are a lot of things that are considered normal that, looking back at those old Disney movies, you go — okay, in the context of the time, that's where we were. But by today's standards, that's not actually acceptable. And I get that this rubs some people the wrong way, but ultimately society dictates progress to an extent, so we have to roll with it — which doesn't always make me the happiest, because I have my own preset of biases too, but at the same time, it forces you to sort yourself out and get better. Learn to understand what's actually going on around you.
Chris Huntingford [00:41:48] It's going to be the same with this, honestly — relying on Excel forever was never going to work either. AI is here to stay, we have to accept that, and it's going to change the way we work, the way we build, the way we deal with people. Which is fine — and going back to your growth comment, this is growth.
Emre [00:42:06] Yeah, exactly, that's exactly how it is. And like you said, especially for people in HR — people dealing with humans on a daily basis, in people-facing roles — how you use AI, how you use it to interact with people, and the role of oversight and training it the right way, is going to matter more than ever. And when I say training it the right way, I don't mean, "Oh, I gave it a bunch of prompts and now we're done" — it's going to be education, like you said, on using AI responsibly, and, as you pointed out, sustainably. That's going to be the name of the game going forward. But also — think about the roles this opens up for people. Yes, AI is "going to take my job" — but there was no such thing as a CISO a few years ago, no such thing as a sustainability expert, no such thing as a red-teamer a year ago. Now I've got a new role as director of AI — what does that even mean? Nobody fully knows yet, but these roles are changing, these roles are coming into being. Another example: the idea that Excel would take everyone's job as an accountant — and yet we have more accountants than we did years ago. It's a shift, not a disappearance — we're just moving to another layer, using the technology we're given. We have to remember — and I'm going to quote Trevor Noah here — we have to start protecting people, not jobs.
Chris Huntingford [00:43:53] As we define what becomes the default for the next generation, which we are doing right now, we need to stop focusing on the job someone holds and instead focus on the person — skilling them, educating them, and finding them a role in the new world. That's what I'm genuinely passionate about: how do we enable the person, rather than the specific thing that person happens to be doing right now.
Emre [00:44:19] I love that. Chris, I don't think there's a better note to end the show on — that's the perfect full stop to this conversation. Thank you so much for coming on the show. Where can people find you? What are you up to recently?
Chris Huntingford [00:44:39] Best place is probably LinkedIn — linkedin.com/in/chrishuntingford, please hit me up there, I'm pretty responsive. Twitter is @ChrisHuntingford. Those are the two main platforms, you can find everything else from there. I welcome messages from everyone — if I can learn something, or share something I've learned, I'd love to. Coming up, I've got two or three conferences: Days of Knowledge in Vienna next week, then Microsoft Ignite in Chicago, and then European SharePoint Conference in Stockholm. So I'll be at those, hassling everyone and talking about responsible AI. I'm excited about it.
Emre [00:45:26] Awesome. And if you're listening to us on iTunes, Spotify, or wherever you find podcasts, leave us a like, subscribe, it really helps the show out. And if you're looking for the ultimate performance management solution in the Microsoft ecosystem, you can always try Teamflect for free by clicking the link in the description. This has been the Team Check-In. Thank you all so much — have a great day! Bye-bye, everyone!
.svg)