#3 - 10 Mar 2026

Episode 3: Talking to AI, How Buying Has Changed & the Birthday Gift Nobody Expected

Kieron's back from a whirlwind week — Glasgow, a college hackathon, a delayed flight, and Tunnock's teacakes. Neil's been holding the fort. Between them, they've got plenty to talk about.

This week the lads get into some genuinely meaty topics:

Talking to AI — are we doing it wrong? A simple conversation about how people interact with AI opens up a fascinating rabbit hole. From a BBC documentary about an AI girlfriend encouraging a man to break into Windsor Castle, to whether AI is making us all a bit... dimmer. Like calculators did for mental maths. Thought-provoking stuff.

The way people buy AI has changed Six months ago, organisations would see a demo and just go for it. Now? Committees, multiple demos, procurement processes. Is AI growing up and becoming a mainstream IT purchase — or are people just getting more nervous? Kieron shares his observations from the frontline.

Product of the week 🎵 (imagine your own jingle) Two new KnowledgeFlow features that are genuinely impressive — safeguarding alerts that flag high-risk phrases in real time, and "Excel Stitching" — a tool that takes multiple spreadsheets and lets you interrogate all of them at once in plain English. Could this spell trouble for traditional BI dashboards? They think so.

From the road Kieron's report from Scotland's Housing Festival — standing room only, a wandering microphone, and the mould demo he forgot to do (you'll have to tune in to understand why that matters).

Oh — and the mystery birthday gift from last week? Duck fat and goose fat. The roasties were, apparently, a triumph.

Neil's off snowboarding next week — tune in to find out who's keeping Kieron company on the podcast bar stool.

Transcript

unknown 0:02

Right.

SPEAKER_02 0:02

Should we get this pantomime horse of a podcast underway for week three?

unknown 0:07

Right.

SPEAKER_02 0:09

Hello. Stop messing around and buckle up because uh we've got lots to talk about this week. Um I've got um three topics to talk about. Um one is um talking to AI and what that really means to have had some interesting conversations this week. Uh second is really around customers and how they're buying AI. And I've noticed a change over the last three three months or so, which has only just kind of really occurred to me this week. And then the third topic was really to hear about your exciting week because you have had the busiest week ever. Well, I don't know about that, but it's been pretty old busy. So maybe you should share with our listener uh what you've been up to this week.

SPEAKER_03 0:54

Well, I'm happy to start there, and um thank you. And first I should thank you for your birthday gift.

SPEAKER_04 1:00

Oh, yes.

SPEAKER_03 1:01

Uh for last week. Perhaps we should wait till the end. You've got to stay till the end to find out what it actually was. Okay. Yeah, that's right. Do that if you want. We could pretend to be pretty funny. Right. My week. Yeah, my yeah, very, very busy week. Um, and today is nice to be sort of back in the office effectively, because I've been running around the place. It started out in Glasgow, uh, which was great fun. Uh, as you know, our Tony, our uh product head of product delivery. Um I never really know what his title is.

SPEAKER_02 1:35

I never really know what his title is either, so don't worry about it.

SPEAKER_03 1:38

He's like you know, he's the guy that kind of makes it all happen in the background and talks to the uh tech team and talks to customers. Wonderful. Um, but he took me on a little tour of Glasgow when I was up there and I was speaking at Scotland's Housing Festival. So that is now I have spoken, it's a chartered institute of housing event. Um, and I have now spoken. I started out speaking at their All Ireland event a year and a bit ago. Uh, and then they said that was okay, well done. You can now speak at our Welsh event, and I was like, great, that's brilliant. And they said, Right, okay, well done. You can now speak at the English event. Uh so I got to do that in Liverpool in September, and then finally, the big time, we made it to Glasgow, which Tony was very delighted about. So um, yeah, that was a really good session, uh, and I can say a bit more about it. And then I came back from then straight into uh West London College, where we've been for two days with our FEAI hat on. I'll focus our partnership there on building products for FE colleges, um, and which is really, really interesting because the as you know, knowledge flow as a platform is brilliant at being your kind of one-stop AI for your staff that's auditable and secure. Uh, but now we're building more and more niche tools for those for those organizations, for the sectors. Uh, and as I was sharing earlier with our housing in a housing conversation, the model which I'm loving that we're doing is we'll work with a college or a housing association for free to build the thing that works for them, as long as they're happy that we can then push that thing to all the other customers in that same sector. So it's kind of a really nice sector-led approach that we're taking. And the West London College thing was brilliant. We had we started with the executive team, the entire executive team and the CEO there um was very, I think, brilliantly astute that you can't just put a bit of AI and hang it around the place and hope, as we uh would say. As we know. You need to manage it and make sure that people feel confident and make sure that we're you know pushing that out in a managed uh change programme, really. So under her kind of uh request, really, we ran these two days of hackathons, which started with the exec team, where we really spent most of the time on AI implementation strategy and have got now some really good kind of at least the building blocks of that, with the full exec team involved and on board and and with that, which is brilliant. And then we did two-hour sessions with each of their kind of key teams. Um, it was an open invite to them to bring everyone in their team or just them and a couple of colleagues or whatever they wanted to do. Uh, so we went to HR, teaching and learning, marketing, uh, flexible learning, which is their distance learning, kind of work-based learning offer. Um, gosh, and others that I'm forgetting, not doing service to. So, how many did you have in the room? Uh up to variously up to probably 10, 12 people in the in the biggest ones and two or three in some of the smaller ones.

SPEAKER_04 4:40

Okay.

SPEAKER_03 4:40

Um, but brilliant, because what we did was straight into a kind of really open, relaxed conversation about what they do and where their challenges are and how we might be able to do some things. And then brilliantly, as uh, I brought I had I had the the whole leading AI team kind of on standby available to join us, and we got Donald in our CTO uh to talk about a really kind of challenging thing, and uh, and it was brilliant just to bring him up on the big screen. He was in about 10 minutes. I kind of briefed him for about 30 seconds on what it was we're talking about. He asked some questions, there was a discussion in the room, and we've gone away with a kind of clarity on what we can do. We'd there's some other work to do. And then brought Ben in later in the afternoon, which was brilliant to talk about a marking, an assessment tool that what would just save huge amounts of time in in kind of cutting through having to having to kind of check off everything in a in a student's work. So really, really good. So, yeah, we I think it's brilliant and it set the college up for I think a really interesting next kind of few months of getting AI into the hands of all of their staff, and I think is a model for the sector because it really that that commitment and the progress that we made with that commitment, I think it was just tremendous. So that of course it's like our whole world has always been, it's now about follow-up. And will we, you know, will they will we get it done? And so we're gonna keep that momentum up. That's the next, that's the next part.

SPEAKER_02 6:11

It was uh I was chatting to Donald last night and he tells a story slightly differently, which was which was very funny. He said, I took a phone call from Kira and he said, Have you got a minute? And I went, Yeah. And he said, Right, and he turned me, he turned it right now. You're on the big screen and you're in front of all of these people. No pressure. So he tells it slightly differently. But I think the thing about that that I really liked was that whole kind of uh we've got this problem. Um, this is what we're trying to do. Can we actually do it? I don't know, let's get on and find a way to scope it out and then create something, draft something. And I think that whole kind of hackathon approach of uh, well, let's build it and let's give it a go is just so powerful, especially when people are skeptical about using AI. Uh it's that whole kind of well, here it is, prove it. You tell me what works, tell me what doesn't, let's fix it, let's adjust it, let's tweak the settings, let's try and make it for you there and then. And and um uh I put out a a piece this morning which was along the lines of you know, other consultancies will show you the demo. You tell us what your problems are and we'll build it for you, and you can tell us whether it works or not. So that's a really different approach, which uh uh I yeah, I really enjoyed out. So I'm glad they uh I'm glad the team were all were all super helpful with that. So it sounds like a really good, a really good session with those, with uh with that coach.

SPEAKER_03 7:30

Fantastic, brilliant, and we want to do more of them now, so it's really yeah, I think that is the way to really drive a good change program, as you know. That's our background, that's what we really know and love, and yeah, that works really well.

SPEAKER_02 7:42

Yeah, and just that that kind of that leads into one of the topics that I had that which was the talking to AI. I was chatting with somebody who, you know, bright person, they'd used tools, and but they said, I don't really know how to talk to AI. And talk was the word that they used. So I just made me think about it. And I said, Well, I did the usual thing of well, you need to treat it like a competition, it's not like Google search and you kind of need to try and have a backwards and forwards and challenge it, don't just accept what it says, go back and push it and prod it and and ask it to critique, not to just tell you what the answer is. And um and he said, Yeah, I know all that, but I just it feels awkward to me. And um somebody else mentioned to me um a piece on the BBC, which you might have seen with uh I think it was Dr. Hannah Fry about people whether people talk to AI, and I haven't actually seen it, but that sounded like a really interesting programme that I should I should go and watch. Did you did you see that one?

SPEAKER_03 8:34

I did, yeah, yeah. I saw uh AI Confidential and it was uh Is that what it was called? Yeah, yeah. I think the the kind of um I guess the hook part was there was a a guy that got into Windsor Castle with a crossbow in 2023, um, and he was going to kill the Queen, and apparently his AI girlfriend had encouraged him to, and they they were looking at the uh the transcripts of those conversations that he'd had. Um but the sort of wider point is what does it mean for us as humans to have this thing that can do amazing stuff and you know, but is ultimately in this case pretty sycophantic, really, isn't it? It encourages you along whatever route you want to go down. Uh so it was an early, it was an early kind of pre-Chat GPT AI tool, really. But um, yeah, it was kind of interesting. I mean, it's interesting. Uh I don't think it I think the that wider question about what it means for our psychology and not just in that kind of relationship, just and I've asked quite a lot about uh sort of challenged, I guess, quite a lot about if we're building all these tools that are kind of doing the thinking, what does that mean for people and humans in that role in the future? And you know, on the most basic level, a policy assistant that already gives you the answers and doesn't need you to know and interpret, you can it will interpret the policy for the requirement you've got. Is that a good or bad thing? And I mean it's a thing and it's a definite fact, is my view. And I think interestingly, as you know, we um in our social worker tools, what we see quite often is a pattern where people will ask more. They'll go, I didn't know that, tell me more about that. So they're educating themselves via the AI response that they just received. So there is some evidence that it's not about making you more stupid, it's about it can be used to you know add your curiosity to satisfy your curiosity. Yeah, but it is a concern, I think, for everybody. But the reality is it's a and well, here's the thing, and I never shared this on stage, but it's I was asked this question at the Scotland event, it's a similar thing. What does it mean for us? We're all gonna get stupid. If you think about what calculators have done for most people's maths, you do kind of think, well, maybe there may well be something in this. So I think guarding it, yeah, go on.

SPEAKER_02 10:52

There were two other things that came that came to mind when I was uh thinking about it. One was the um and it was related to actually use of the AI tools in education, in school specifically. And it and they the the person said, uh can your AI see what students have written down? And uh uh I said yes, um uh it is in the um safeguarding um administration tool. And um, you know, that's super important being able to spot those uh those areas. Um not everybody has that switched on or rolled out, but I think it's one of those areas that probably more people should than not. I think it's it's uh a really interesting debate. And then there's a second, which was uh something you raised when we were talking earlier about um uh AI knowing you and having to start again every single time. And um how do you create a an A, how do you how do you create an AI that knows so that you don't have to say don't write with bullet points every time or don't write with um you know, don't use the M-dash or whatever else people get muppedy about on using chat GPT.

SPEAKER_03 12:04

Yeah, you're right on your the amnesiac AI that you come back to and it's like fresh again every time. And yeah, some of our tools are like that. And yeah, I was just chatting with Donald about this uh on Claude, Claude have uh got an option there, Claude Co-work, which we can't use, of course, because it's not secure enough, but where they've got a way of remembering what you wanted to about you, and I think that we could build that into knowledge flow. I'm quite interested in that same piece that would enable you to well first you pick up from your conversations about some of the your likes and dislikes, but also you you could actually go in and say, here's my writing style, here's some whatever, if you wanted to. Um yeah, interesting. I think I'm it's it's an essential, really, isn't it? Now I think I use perplexity for search, as you know, I think, and um that has a memory built into it, which is quite handy. I don't have to tell it about which MacBook I've got if I'm asking about that, and indeed I don't have to tell it about knowledge flow if I'm asking about that. It knows I run an AI company, but knowledge flow is our platform, and it's quite handy because it will quite often say to me in his answer or say, and this is relevant to you in your in your uh knowledge flow development, which is quite fun.

SPEAKER_02 13:21

Tell me more about Glasgow then. So um I used to live in Glasgow, I don't know if you knew that. Um used to hang out in the West End. That was that was great, it was a great party town.

SPEAKER_03 13:30

Um I enjoyed the crack. I don't suppose it's called the crack there, is it?

SPEAKER_02 13:34

It's probably something else, but I'm not even gonna vaguely pronounce it because uh I'll end up in as much trouble as you. I it's a long time since I lived there, but uh yeah, it was.

SPEAKER_03 13:44

Tony was telling introduced me to what he described as trad music, which was five people with an accordion and one person with a guitar, and someone with some bagpipes. Uh which was of course they had bagpipes. It was it was gonna be. This is my bias. I'm like, I'm here I am, like Chat GPT now. Yeah, exactly. Give me a picture of an accordion player. Yeah, you've just got a bias game.

SPEAKER_02 14:16

So, what was the uh what was the best thing about uh um Glasgow and what was the um what was the key takeaway that you think people would have would have taken from your session?

SPEAKER_03 14:27

Oh, that's tricky, isn't it? I my story arc uh for my session, which is probably the best thing, of course. But no, I'm sure um the Tunnocks, the Tunnocks tea cakes were probably the best thing, let's be clear. Um and this and a qu a close second was my AI for for strat AI strategy. My story arc was the usual kind of start of AI's here, it's gonna transform things, and here's some things you need to know about AI in order to kind of help you as a senior leader think about where that fits in your organization.

SPEAKER_04 15:01

Yeah.

SPEAKER_03 15:01

I then did a little bit of a let me show you a couple of things. Let's not just talk about how AI is gonna help. Let me show you some of the work we're doing with other housing associations. So I did a quick flick-through demo of those and then got on to so what do you need to do, which is strategy? So, which is around, as we know and have said many times, is um uh it's AI and AI implementation is not a technology problem, it is a skills and culture challenge, it's a change management challenge. Um, so it's sort of one element of your strategy, and I'll give you the probably the one that needs doing pretty urgently is is is that skills and culture. Then it's the products and technology and priorities, what are you actually trying to fix with what? And then you've got data, which is really important for everybody to get on the road with, and finally governance, which is if you're not governing the thing and having who decides what AI we're taking, who decides and and indeed what level of sign-off is appropriate. So is it you know, when do you need the board to sign off that you're going to try some new AI? Because if you're using kind of you know all of your data and allowing it to get towards decision making, which is the dangerous world, you probably do need board. There's enough risk there that you certainly want ethical check questions versus can I summarize this document, which is a personal decision, probably. I'm right, and there'll be different hierarchies of that. Um, and in governance, as I always sort of say to people is thinking about you know how fast or slow are you going, do you have red lines that you want to introduce right now to your organization? Such as in some I know some independent schools that say we are not using AI for assessment for marking, um, because they feel that that's what parents are paying for, and they're paying for humans to do that. I don't agree with them. I think AI is better than humans at that most of the time, but it's not my it doesn't make a difference what I say. As I as I say when I'm on stage, it's not as irrelevant what I think. That's a governance and uh a governance decision. Um the only thing I always say to people is if you are gonna do any of those red line things, just make sure you review them in a year at least, maybe sooner, because the reality is AI is catching up all the time. So yeah, so Scotland was great, and um and I have to say also I so I flew up because it's like five hours and forty minutes from Houston to Glasgow on the train, and that's a long, tong time, isn't it? So I flew up and it was cheaper, in fact, it was quite a lot cheaper to fly. And uh on the way up there, it was a bit delayed. They did a thing with a list, we were on board the plane, and the captain said um he said sorry sorry about the delay. Uh we we realised we did something, we packed the bags wrong or something, so they had to take them all off, re-scan them all, and put them all back on, which they did do in about in less than 10 minutes, which is pretty impressive but annoying. So I was like, I'm going by train next time. Because that's the trouble with the flying, as you know, is that you've got bloody taxi tube waiting in line to get to get searched, an hour maybe knocking around in the airport, half an hour waiting in a queue to board, sitting on a plane that you can't really get your laptop out easily. All wasted time. So I and anyway, and I flew back and they took off early um and landed half an hour early into Heathrow. So I was kind of like, oh, now you've put me back in and do a flying train back.

SPEAKER_02 18:21

Yeah, and I have to say, sometimes it takes more than uh five hours and 40 minutes to get via train to uh okay. Sometimes there are a few challenges. I know that west coast line very well. So I shouldn't say okay. Now, did you do did you do two things on uh stage in Glasgow? I want to know, did you um input the picture of mould and get it to analyse it? You forgot.

SPEAKER_03 18:48

Forgot. It's so annoying. The one thing I mean the the the biggest challenge I have on stage is I've got so much I could share. And I do a few rabbit holes, partly on purpose for authenticity, and partly because I get a bit carried away and decide to share something else. But my when I was doing the demo, I was very conscious I needed to move at pace, so I forgot to do because as you know, what the thing is amazing, you can upload a picture of mould to the housing repairs assistant, and it will look at that, tell you what it is, categorize it as the level of risk that mould is, and then give you the what you need to do next.

SPEAKER_04 19:25

Yeah.

SPEAKER_03 19:26

Incredible. So you could have a world where a tenant via an app uploads the picture of the mould and it could go straight through into a repair, no human touching it if you wanted it to, or or not. It could go through you know lots of approval steps if you wanted it to to.

SPEAKER_02 19:42

Sounds brilliant, and you forgot to do it.

SPEAKER_03 19:44

Fabulous. I know, I know, I know. It was so many things. I yeah, as always, it's my it is my challenge, but it as you said when we talked about it. I for me, when I put my slides up, I don't really know what I'm gonna say. I got loads of notes. I wrote loads of notes to sort of for confidence, I guess, the night before in the hotel, and then in the morning, and then I said to Tony, who was Tony came to the session, which is great to have him there, and I said to him, You watch what I'll do. Um you just watch me. My notes are all there, right, on the lecture, and I don't put them in my hand. As soon as I start, I'll walk off. And I did, and I find myself standing on the other side of the stage, like, and I'm like, oh no. And then the next slide comes up, and I'm like, Okay, there's that one, brilliant. I know the slide, I know I obviously put them together, but I don't quite know what I'm gonna say and for how long on each of them.

SPEAKER_02 20:33

He did say you wandered around like a TV presenter in uh a talk show or your microphone.

SPEAKER_03 20:39

Did he?

SPEAKER_02 20:40

You always wanted to be on TV.

SPEAKER_03 20:42

Oh, there you go. Yeah, I did walk into the audience when uh there was a question, so I yeah, I wandered off with it. So that's a long way. Yeah. Very good. It was really good. And um, yeah, we got a few interesting conversations straight afterwards. Um we were asking people if they want to kind of pilot stuff with us, then let us know. Uh and we had, I think what, nine differ nine of them that have got in contact via the website for more info. So it's all yeah, that feels like a good place to be. Yeah. Did you show the Excel stuff then? Uh I nearly I can't remember, probably. Um I have though all week been talking about it in our new product. So this is pretty much.

SPEAKER_01 21:23

This is why I was asking whether you presented it. You'd be so excited about it earlier in the week. I thought you must have you must have done it. But it yeah, now I'm worried that given that you forgot the moldy bit.

SPEAKER_03 21:33

Here's the thing you've got to know about when I stand on stage at any of these events is I try to limit talking about our products because they're not, you know, I want to give people lots of value and not be all about what we can do. Sure. I want to show them what's possible, and that's part of that is the case. But um I don't know. Ask Tony if I shared anything. Probably.

SPEAKER_02 21:58

He won't be able to he won't be able to remember. Has stuffed himself full of tonics of tea cakes and his mouth full.

SPEAKER_03 22:06

But products of the week, though, we need a jingle, really, don't we, for products of the week? Ha ha brilliant. We'll get one. There you go, there you go. But insert jingle here. I tell you what, I saw uh Stuart Francis, as he called a comedian. Yeah. Um, I saw him speaking and he was talking about Brexit. And he said, the most annoying thing about Brexit. He said is uh he said, I release a DVD and um I if I put music on it now I have to go and get the approval in each jurisdiction. And he said it's a nightmare. So he said, so imagine what I want you to do now is imagine some music playing in your minds as I speak about the next bit. I mean it's brilliant. So if you imagine your own jingle now, listen. All right, I've got it. It's product of the week. So um you talked about the safeguarding, we've got safeguarding and Excel stitching, as we call it, which is as usual coming up with our own names for stuff, to make it more simple and easy to understand than the way the rest of the world works. Safeguarding's really cool, you talked about it before. That's about triggering we get automatic alerts, uh, and uh customers get automatic alerts if high-risk phrases are entered via their knowledge flow platform. So it's it's essential for any student-facing tools, the FE require it, you've got to have it, and indeed there is a real challenge for schools and colleges, and I say this to them a lot, is we can trigger the alert to you. And if you get that alert at midnight and do nothing with it, and let's say the worst happens, you knew at midnight there was a problem, but you didn't have staff in there, and that's a new problem for most colleges and schools, as you don't have that, it doesn't exist really at the moment. So there is some real I mean those workflow wider problems than than sort of the AI's challenge, but it is worth noting. But yeah, yeah, the safeguarding thing's really good because it just gives that extra layer of protection um for everybody involved, which is great. Uh but Excel Excel stitching, so knowledge flow can has for a while been able to get an Excel and do amazing things with it. Um we've now made that uh well, I say we, I mean this is all Donald's work. I'm just basking in the glory as usual. Chief technical wizard as I've as I've titled them. CTW. But so so it can take sort of unlimited Excel files and multiple sheets, and it handles all of that. It creates an SQL mirror matching SQL database. But where it gets clever is it will automatically create a relational database between different sheets or indeed different Excel uploads and match automatically on what it can see as a common identifier. If there isn't one, obviously it cannot. But it's even cleverer than simply the sort of normally with a common identifier, you have to have an exact title, exactly the same. But there's some good AI stuff going on that means that it will infer a little bit, so you need to check, but it will infer if it looks like it's the same thing, just called something different in that system. So student ID might be called student number or student reference or reference. Um, but if they look the same, then it can do a uh a relational database. And the reason that matters, as you know, is that you've suddenly got this wonderful world where you can upload three, four, five different data sets and ask questions in natural language across all of that. And I I mean that excites me enormously because as you know that I keep saying as you know, I'm gonna stop saying that. Um things like Snowflake, the data warehouses, the data lakes that are um big and uh people spend a lot of time and money trying to get their data into there. I keep on wondering. It is no doubt easier for if all of your data is well organized and categorized in a warehouse or like a because you're pointing the AI at one place and you've already in the background done the work to kind of clean it up and join it up. So it's easier, but that's expensive, and it's not easy to get to that point. AI can take it straight from the database or the spreadsheet. Let's be clear, you don't necessarily need to even have it in a in a nice structured database. So I I kind of the future of that is I think really interesting, and I'm not surprised to see the stock value of some of those companies not going in the right direction, or not the way their shareholders would want it to go anyway. Because I yeah.

SPEAKER_02 26:43

I got a um I got uh an email this week from a council saying, uh, would you be interested in bidding for a uh BI project? Um we want to put BI on top of some data set and create dashboards for various aspects of the work that we're doing. And I was like, I don't think I don't think there's a future in that. It's like actually, but there's a much easier way of doing that, and it's not the way that you've set out this project. It would be much better to just take that data, stick it into um uh stick it into knowledge flow or or other tools are available for the uh for people for the for the one listener, um uh but uh then using natural language querying on top to actually get the the information that you want. And you can use visualizations then to create pie charts, graphs, trend analysis, whatever you want, but creating you know a hardwired uh BI powered um solution is just old-fashioned thinking.

SPEAKER_03 27:51

It is it's 2024 speaking. Yeah, yeah, yeah, that's right. Uh, infographic is you can ask it for an infographic, which for most humans is better than a dashboard. Yeah. And it because it just explains it so much nicer, normally, nicer is one wrong word, but yeah, I think that is absolutely the future of it.

SPEAKER_02 28:14

But that kind of links to what the one of the things I said right at the start, which was one of my topics, was I I I think the way that people are buying AI has changed. And I've only just realized it this week. And um before people used to go, oh yeah, let's give it a go, let's give it a let's run a trial, or let's just, yeah, it's just it's um it's cheap enough that we can just put it under the procurement radar and we can just get on with it. Let's put the guardrails in place for data and GDPR and you know, all of that contractual stuff that we actually have to do. But one of the things I've noticed this year, and so we're only in you know month three, we're only at the start of month three, really, um, but there's more committees, there's many more demos. We used to do one demo and people go, yep, like it, off we go. Now we're doing at least two, more often three, sometimes four. And is it just because people can't get all the decision makers on a on a team's call at the same time? Or is there more to it than that? And and in a previous life, I used to work um in an organization selling um risk management software for clinical trials. So you can imagine that that was quite tricky. You know, we would expect eight or nine different groups for it to go through, because you had to go through quality assurance operations, etc. etc. And um uh but with a simple rag solution, uh I'll give you an example. Um I chatted to the some uh GPs and they said, is it a class one medical device? I was like, what? It's like is it going to be giving medical advice? It's like, no, we're talking about policy body so that your people do the right things and don't say the wrong things. You know, we're not talking about you know medical diagnosis. I don't want to get into that, but if you wanted to, then that is a very different thing. But don't assume the worst case scenario. Let's just start with some simple things like you know, how are you how are you answering complaint letters? Uh because here's the other thing. One of the things they said to me, and you must have seen this in housing, because I uh um because I don't deal with housing, I don't really know. But one of the things they said to me was that the complaint, the quality in inverted commas of uh complaint letters have gone through the roof because um people are using chat GPT and others to uh write high high quality complaint letters. And if the response that they get back isn't high quality, then all they do is just stick it back into chat GPT and say, What do you think of that? It comes out rubbish, go back and ask it. So they end up in this kind of escalating cycle. So stuff that used to be easy now isn't. And so I was like, Well, you could just that's not a class one medical device, don't make it harder than it is. That's simply answering um uh different categories of um complaint, some of which the AI can't handle. Absolutely, it's going to need your GP to answer that, if not the practice, the practice manager or the GP needs to answer those, these ones. But these simple ones here, absolutely you can sort those out. So really interesting that this it feels like AI, whereas it used to be sort of a bit kind of niche and a bit sort of projecty, now is becoming much of a more mainstream IT purchase. I don't know whether that's just my observations in the conversations I've been having. But that was that was one of my revelations from this week.

SPEAKER_03 31:41

Yeah, interesting, really interesting. And and I think I I mean I recognise all of that, and often we'll do a demo and there people are blown away by it, and rightly so, because it is very impressive, and not just because it's a bit of AI, but because these are niche products designed for their sector, so they can see things that they just couldn't do with any of the other products very interesting.

SPEAKER_00 32:02

Taking pictures of mold and telling you the category of quality and what they need to do about it, and then routing it through to the repairs people who then can get the right stuff and fix it. Exactly. Stuff they didn't even know existed just as a one told them.

SPEAKER_03 32:16

There'd be an example, but you know, there are others. But then often what happens is they often like that, then have to take three steps back because they realise AI can do lots of cool stuff. Now it's like, oh, so I I thought it was just a you know thing I was gonna see and maybe buy for me, and then you see how important and impactful it could be, and I think it ends up being a bigger conversation, uh, which is a real challenge in our world, as you know, because we we need to bring in enough uh new customers each month to make sure that we're able to grow and keep on doing what we're doing. So trying to force people to go a bit quicker. I mean, public public services aren't known for their uh pacey buying at the best of times, let alone when it's something novel. That's a great segue.

SPEAKER_02 33:01

I I have I had two lessons from this week. Last week you asked me about my lessons learned. Yes, one of my lessons was speed kills. And um uh uh we used to talk about that when we were playing rugby. If you couldn't catch the man with the ball, you had no chance, they were going to score every time. If they got it to their fastest player, they were always weird. Speed kills. And um, there are other analogies um that that you can use in life, but um I'm gathering you weren't the fastest player at any point, so I don't think this was applying to you. You are so rude. Uh there was one time I scored a try, and uh my wife was came up to the pitch afterwards, and she was chatting to one of my friends. She said, Oh, it's Neil playing, where's he playing? He's playing, oh, he's playing out on the wing. He's just scored, and I can't tell you the words that she used, but one of them was she she didn't believe that I'd scored. Outrageous. Anyway, come back to my point about speed kills with AI, speed kills, two things. One is things are moving so fast, and uh, you know, it's it's only been a week or so since Claude did an upgrade. This morning, I I saw that uh um ChatGPT 5.4 has been announced, so the models are just moving really so quickly. And then at the other end, speed kills because people go so slow. I've had a number of conversations this week where people say, Are we just not ready? or or the team aren't ready, or or or it doesn't, it doesn't quite do this, or you know, I wanted toast, but can it do eggs as well? And is that just actually the stalling piece we I know find frustrating, but yeah, that we've got two examples of speed. Some people are moving so incredibly fast, automating things, automating workflows, making stuff happen, taking away those drudgery and um difficult tasks. And at the other end, people still stalling because it's I don't know why. Uh I I I've I've had lots of conversations with people about it. And uh lots of people talk about fear of AI. I'm not sure it is fear of AI. I think it is that kind of almost, I'm too busy. Uh so messing up the status quo is too hard for me, so I don't really want to do anything. But yeah, speed kills was one of my lessons learned in both going too fast and going too slow.

SPEAKER_03 35:22

Yeah, how interesting. I think I mean uh we see it all the time, it's very frustrating. My biggest bug bit and that and getting out to sharing enough sharing with enough people what we've got because knowledge flow as a platform is absolutely tremendous. I mean, I keep up with AI constantly, and yeah, I mean there's open claw and clawed co-work are incredibly impressive tools. Knowledge flow is not very far behind and is a secure and that which it changes everything and gives you consistency across your whole organization, doesn't need a load of AI skills for your staff to use it, but can equally accommodate people that want to you know create brand new amazing things that and test stuff out. So it's just I wish we could I kind of want to just kind of grab every CEO and just have a look at it and understand what it could do for your business. And as you know, the passion really at the heart of it all is I want to make public services, and I know you do too. I want to improve public services through AI because it can hit so many of those entrenched problems that have forever caused trouble in public services. And I want to do that, and we are deeply inexpensive, and it's frustrating as hell when we just can't get in front of people, or you know, when they go, Oh, yeah, that's a bit more expensive than I was thinking. Well, you're thinking what you think is a fibre, you know, how much more we we are not expensive, we are not by any measure of the word. Yeah, even my kids wouldn't think we're expensive, and they think yeah, they're profligate, aren't they?

SPEAKER_02 36:58

Anyway, speaking of kids, did you have them around for Sunday dinner?

SPEAKER_03 37:02

Oh yes, coming to this uh so what did I get for birthday? I got a kilogram of goose fat and a kilogram of duck fat. It's quite possibly the weirdest present I've ever received. When I got them out, I said, what are they? And then and then and then it all came because you've given me a few clues, hadn't you, to uh that it would I'd I'd like it more by Sunday afternoon than I would when I got it. And I was and that's when Donald said, Have you got a foot spa? It's swimming. So but thank you very much. I used duck fat, made duck fat roast potatoes. They were a big success, and we were deleting them on Tuesday, and they were delicious. Oh, excellent. Very good.

SPEAKER_02 37:48

And you're off snowboarding now, aren't you? I am, but that does mean that you're gonna need to find a new partner in podcast crime for next week.

unknown 37:57

Wow.

SPEAKER_02 37:57

So you better give that some thought and uh and and try and find uh someone who's who's prepared to to talk nonsense with you for uh for 45 minutes.

SPEAKER_03 38:05

Well maybe you should dial in from the slopes and um I thought about it. Perhaps not.

SPEAKER_02 38:13

But yeah, that's right. Yeah, I'm not sure you want to see me in that kind of uh I was gonna say stit. I do mean state, yeah, yeah, yeah. No, I'm not doing that. No, it's a holiday. I'm taking some time off. Um, but uh yeah, you need to find a new uh new partner in podcast crime. And then maybe our listener won't fall asleep if he's um if he's got someone.

SPEAKER_03 38:35

Who knows? We'll see if we can see if we can find other guest.

SPEAKER_02 38:39

Are you talking about podcasting, or is there something else you need to tell me?

SPEAKER_03 38:44

Sorry, you had to find out through this uh podcast.

SPEAKER_02 38:47

I'm gonna find out I've been fired through the medium of podcast. I wonder if I'll be the first person.

SPEAKER_03 38:53

There you go. Right, though. Well, listen, have an amazing time on the slopes. I'm not at all jealous, and um, yeah, well, we'll look forward to your return in two weeks on our podcast.

SPEAKER_02 39:03

Excellent. All right, fella. Well, I will speak to you when I'm back. In the meantime, have a great weekend. Uh try goose fat um rosties this Sunday and uh let us know how you're doing.

SPEAKER_03 39:16

I will too. Thank you, buddy. Thank you.