S2. Ep7 - Does AI Actually Save You Time? (Harvard Says… Not Really)
Does AI actually save you time, or is it just creating more work?
In this episode, Katie and Noel dig into a Harvard Business Review report that found AI might not be the time-saver we all think it is. They share their own experiences of falling into the AI productivity trap — from filling saved time with more tasks, to taking on responsibilities that were never yours, to that "one last prompt" habit that means you never truly switch off.
They also chat about the new Kimi 2.5 model from China and its agent swarm feature, why automations might be the real answer to the productivity problem, and Noel's hilarious battle with ChatGPT over whether to walk or drive to the car wash. Spoiler: Claude got it in one.
How to find us:
Join our membership over on Skool, where we support you on your AI and automation journey. We share exclusive content in the membership that shows you the automations we talks about in action how to build them. Find out more about the AI Business Club here.
We have a free LinkedIn group (AI Automations For Business), the group is open to all.
New for 2026, you can also find us on Substack, click here to subscribe and get all the latest news and updates from us.
If you would like dedicated help with your automations or would like us to build them for you then you can find our agency at makeautomations.ai
-
Katie (00:26)
Hello, welcome back to another episode. Hi, hello, I'm Katie. And as always, I have Noel here with me. How you doing, Noel?
Noel (00:35)
Yeah, doing great as always, thank you.
Katie (00:37)
Good to hear, good to hear. I feel like it hasn't been that much of a crazy week in the world of AI, but actually, if I start to think about it, very often, so Noel and I both work at home, but in different offices, and we come downstairs to have lunch together, and it's usually something business related. And I'm just thinking back to our lunchtime kitchen chats and how much of that has been occupied by AI and actually quite a lot since our last podcast episode.
Noel (01:21)
Yeah, there's been a fair few things going on, hasn't there?
Katie (01:25)
Yeah, so I was obviously very excited because I thought I had found something AI related that you might not have heard of. And I was really excited to tell you. But of course, Noel's heard about it. So I kind of burst into Noel's office after I'd seen a TikTok and I was like, Noel, have you heard of Kimi K2.5?
Noel (01:38)
Of course I have.
Katie (01:58)
I was like, darn it. Listeners, he did not share that information with me.
Noel (01:58)
Well it came out at the wrong time because it got lost in all the noise of OpenClaw and Claudebot and Maltbook and all that sort of stuff. Yeah, poor timing on their part. But yeah, it did get into my YouTube. So I did watch a few bits about it.
Katie (02:16)
Okay, so what can you tell us about it?
Noel (02:28)
So mostly, I guess it's kind of like a normal chatbot like ChatGPT, Claude, that kind of thing. It's got the usual stuff. So reasoning and searching and all those sort of good things. But there are some bits that make it a bit more unique. So they have the ability to build websites. That's a dedicated area within that chat window. So I guess they've trained it a lot more on website design and things like that. Because one thing I find is most vibe coded or AI-created websites, they all kind of look the same. They all use the same component libraries. They use similar colours and yeah, it's quite hard to make something look unique. I've been trying to do that last week.
So yeah, they've got that part in there. And I guess the other part is they've introduced something called an agent swarm, which is really interesting. So you could give it a really complex task to do, and it would then go off and create up to a hundred mini agents that would go off and piece it all together and do little bits on their own and then collate it all back and give you the response. So that's something that doesn't really get done currently. I think you can do it inside Claude Code. I think that's in there, but it's kind of buried a little bit. But this is the first time we've seen it in a ChatGPT style interface, which is pretty cool.
Katie (04:06)
So I guess the question is, would you be willing to use Kimi K2.5 for your business? Because I think that's what everyone wants to know. Would Noel use it?
Noel (04:22)
I've tested it out, of course. I haven't really let it loose on any business tasks just yet. I've got other things I've been tinkering with. It is a model from China. So it is in the realm of DeepSeek and do we trust where the data is going and all that sort of debate comes back round that we had this time last year. So I don't know. I would have to do a bit more digging into the privacy and how data gets stored before I could rubber stamp it and say, yeah, I would happily give my business data to it.
Katie (05:20)
Yeah, okay. All right, well we'll stand by on that one and wait to see what happens. But if you're using it, we would obviously love to hear from you. Do let us know, either in the LinkedIn group or drop us an email. We'll put those details below for you. So yeah, feel free because we always love hearing from you guys.
Noel (05:27)
Absolutely.
Katie (05:42)
Okay, so shall we get into this week's episode, Noel?
Noel (05:47)
Absolutely. Why not?
Katie (05:51)
It's all about AI. Does it reduce work? Does it create work?
Noel (06:01)
Yeah, that is an interesting take because I initially spotted this on my Substack feed and someone had created a post very similar to the title you've just described there. And I was like, really? What's going on here? You know, surely AI should be reducing workload or making you more efficient and not creating more. So yeah, I was like, that's an interesting take. Did a bit more digging on that one to find out a bit more.
Katie (06:34)
Yes. So we've actually got a report that Harvard Business Review did. And I do find their reports quite interesting. They do a lot on all sorts really, but they've done some really good ones on community and things like that before. So yeah, I find they're always well written and easy to read. There's not a lot of jargon and stuff with them. With that in mind, with the Harvard Business Review findings, I think it's quite good to go through the key takeaways from there and have a bit of a chat through them.
Noel (07:02)
Yeah, I guess the first one we talked about is it's not saving time, but it's kind of filling that time saved. So you might have saved 10 minutes, but then you're using AI to fill that 10 minute gap and you're adding more stuff to your plate. I guess that was the first key takeaway.
Katie (07:48)
Yes. So I'm going to give an example of this. When it comes to me creating content, at the moment for my own business, I'm really trying to get into TikTok. I'm doing this 30 day TikTok challenge. So for me, obviously I'm creating a lot of videos. When I first started this challenge, TikTok is a new platform for my business and it is very different to other platforms. It's very different to Instagram in terms of what you have to post and stuff like that. So I was thinking, well, my mate Claude can help me out with this. And actually I was wasting so much time with Claude, whereas I was kind of like, right, I know the content pillars that I want, like I need to do some storytelling content, I need to do some problem aware content. And then I just started writing out my own hooks. And I was like, would this entice me? Yes or no. And the no's, I was just tweaking myself. And then I was able to just kind of do free form with the video. So I had my hook and I knew what I could talk about.
Whereas at the very beginning I was asking Claude not only to write some hooks for me, obviously I gave it all the details, and then it was writing scripts for me and I was trying to be perfect with this script. So then I was having to do god knows how many takes.
Noel (09:30)
Yeah, I could never do that. I could never read a script.
Katie (09:38)
So actually I've saved myself time by just doing it myself.
Noel (09:50)
Yeah, that makes sense. Because you kind of figure out in this AI world what we can and shouldn't use it for. It's like, you can give it a go and then realise, actually, I could be a bit more creative or save a bit of time just by doing it myself.
Katie (10:02)
Yeah, and I still kind of use it to help me with blog posts in terms of like, can you list some bullet points for me? But again, it's me writing it. And actually rather than asking Claude to write it for me and then me having to go through and edit it and making sure it sounds like me and removing things that I wouldn't say or thinking, right, well it said this and now I need to make it a bit more human or add a story into it. I was just thinking, just give me some key points because then I can just write freely and actually it takes so much less time.
Noel (11:00)
Yeah, definitely.
Katie (11:06)
But am I filling that time with other things?
Noel (11:11)
Yeah, I find I do when I do it.
Katie (11:17)
I think when I was really using Claude a lot for my content, I was trying to fill my time with other things and I find myself procrastinating a lot more. So I was either filling my time with other tasks or I was doom scrolling basically. I'm just going to be honest. It's kind of like, oh well I've done that and I've done that and I've done that. Don't know what to do next, haven't got calls for a while. And I would find that actually I was either doom scrolling or just filling it going, oh I could do this, I could do this, I could do this.
Noel (11:50)
Yeah. I find that a lot with Cowork. So Claude Cowork saves me a lot of time. When we process the podcasts and things, all I've got to do is give it the transcript and then that kind of starts the process and uses all the skills. That could take probably about 15 minutes to complete, but yeah, I just then like, I'm sat around twiddling my thumbs thinking, now what do I do? So I open another tab of Claude and I'm like, well, and off I go again.
Katie (12:46)
Yeah. And another thing that the Harvard Business Review said regarding this was about task expansion. So people start doing jobs that weren't theirs because AI makes it feel possible. So a business owner who used to outsource copywriting or basic code, now they think, I'll get Claude or ChatGPT to do it, and suddenly they're juggling extra responsibilities.
Noel (13:26)
Yeah, they've saved that 10 minutes with AI doing that task. And then they've expanded their job role straight away. It's interesting how that works. I mean, I obviously do it. I'm not a coder or programmer. So I end up doing stuff with Claude Code and things like building things that previously would have been impossible for me to do without lots of learning. So yeah, I do get that it does creep in a bit. But at the same time, if you're building something that's going to be genuinely useful for your business, either generating revenue, getting you leads, or it's going to save you time, then why not? But yeah, I can see on scale in a business, it could be quite hard to manage people and making sure that they're staying in their designated roles. They're branching out and all of a sudden doing things outside their remit.
Katie (14:29)
Yeah, they're not taking on more responsibilities and therefore then all of a sudden are not getting paid for it. I can totally see that happening.
Noel (14:39)
Definitely, it's dangerous.
Katie (14:41)
Yeah, I think it's really important, isn't it? It's practising those boundaries. Kind of just say, actually, this isn't part of my responsibility. Especially for freelancers and contractors, I think that's really important.
Noel (14:47)
Yeah, definitely. Don't get carried away.
Katie (15:12)
The next thing that the Harvard Business Review found was that work becomes ambient. You're always on, never quite done. And it was saying about the "quick last prompt before I leave," which apparently is becoming a habit. I'm guessing this is more for those who are employed rather than for business owners.
Noel (15:43)
Yes, but I think you can kind of get stuck in it as well as a business owner, because obviously with these sort of models, you could be sat having lunch and then just think, I'll just ask this question now, or get it to research something for me while I'm cooking dinner or whatever.
Katie (15:50)
Yeah, I'll just get it up and running. By the time it's had a think about it and spat out an answer, it'll be ready for when I'm ready.
Noel (16:15)
Yeah, exactly. I have a very big habit of doing that. Before lunch, if there's anything I think, I could set Claude off to work on this and it'll be done by the time I've had lunch, then yeah, I'll do that. I would never chat to it during break periods. I'm quite good at that. I never chat to it on an evening. Unless there's some sort of trivia question or something, I'm like, is that true?
Katie (16:46)
Okay. And I'm guessing if you're running automations, it almost doesn't feel like work because you're like, well, I'm just going to set this automation off and actually it's just going to run and really it's saving me time.
Noel (17:07)
Yeah, it's interesting how it's starting to leak into people's lives, isn't it?
Katie (17:19)
Yeah, and sometimes actually when I feel like either maybe I'm feeling a bit overwhelmed or there's something that I need to overcome or work out, whereas maybe before I might have had a coach or actually spoken to a human, I'm asking Claude or something like that. And it's kind of like it doesn't almost feel like work but actually it's business related so therefore it is work. And of course it's so easy to do when you are sat on the sofa on an evening or on a weekend. A thought pops into your head and we've just got it at our fingertips, haven't we?
Noel (18:15)
Yeah, I've had ideas that I was just like, oh, I'll put that in there. And then close it and then be like, I'll come back to that on Monday, see what it comes back with. I don't tend to go full in in my spare time. At least I don't think I do. I used to. But yeah, I've learned that boundary.
Katie (18:25)
Yeah. No, I think you've got lots better with that. You used to be terrible for it.
Noel (18:45)
Thank you.
Katie (18:45)
Well done, well done. It's only taken about five years. But I'm happy for you.
So the next thing it mentions is multitasking. Now, I don't really feel like multitasking is a bad thing.
Noel (19:00)
No, I do also fall into this quite a bit as well. So last week I was building a web app and getting ready to show it to a potential customer this week. So I would have Claude Code working away doing that part. And then I got started thinking about websites. Then I had Claude Code somewhere else doing the front end web design. And then I'd be chopping and changing between the two. So yes, I'm creating thousands of lines of code without doing anything, that's great. But it does become difficult to remember and get the context right for each thing. So what did that one do half an hour ago? I haven't looked at it. What did I ask it to do? And it's like, yeah, right. Now I can go again and ask it to do something else and then move back over to the websites and stuff.
Katie (20:21)
So with that multitasking, do you think it actually creates more work for you?
Noel (20:40)
I mean, it saves work in terms of I don't have to sit there and write every line of code. One, I couldn't, but even if I could, I couldn't type that speed if I tried. So yes, it saves that time, but it's also, I guess, going back and reading and checking what it's done and things like that. So that can slow things down a bit. I would spend more time thinking about what it's done, what was I going to do next, I can't remember. So I have to think about that for a bit. And then, right, that was it. And then set it off again. So yeah, you've almost got to get back up to speed with each task as it finishes if you're multitasking.
Katie (21:16)
But does that make you feel productive? Because I'm guessing that might be your outcome. Is it to feel productive or accomplished? What does it actually make you feel like being able to multitask those different AI or automation tasks?
Noel (21:29)
I guess it does make me feel more productive. It gives you that sense of achievement as well. Not only did I get this app ready, but then I created some hero images and I was sending those off to people to say, what sort of designs do you like and that sort of stuff. So I created 11 sample hero images that were all done with AI. That task in itself only took about an hour and a half. Imagine trying to think of a design, code the design. I could never have done that in an hour and a half. So yeah, definitely feel a lot more productive.
Katie (22:07)
Okay. So let's go back through those four main points that the report found and we'll go through and say, do we agree, do we disagree, or maybe what we would do. So, it fills the time you save.
Noel (23:05)
I would agree with that, yes.
Katie (23:07)
I'm also going to agree with that.
Noel (23:11)
Yeah, definitely does, doesn't it? It's too easy, that's the problem.
Katie (23:13)
It's too easy. And then the next one was about more tasks and more responsibilities. So because you think, or someone thinks, I'll get Claude to do it, you feel like you can take on that task or that responsibility. Agree or disagree?
Noel (23:40)
Definitely agree with that.
Katie (23:43)
100% agree with this one.
Noel (23:46)
Yeah, I mean, you could code your own website within Claude in 10 minutes time. That's something you wouldn't normally do, but you could do it.
Katie (23:59)
I still don't want to do it.
Noel (24:01)
No, I know you don't.
Katie (24:05)
You're still trying to make me do stuff, Noel.
Noel (24:05)
Leave it to me.
Katie (24:06)
Okay, so the third one was you always feel on and you don't feel like you can switch off from business. Agree or disagree?
Noel (24:25)
For me personally, I would disagree, but I do agree with the sentiment. Because I've been there, done that and come out the other side, thankfully. So yeah, I could see why people would fall into that trap.
Katie (24:37)
Yeah, I do agree because I'm thinking of business owners in general and kind of blanketing everyone together. I just remember my early years in business and I just didn't have any boundaries. I was just working, working, working. Whereas now I'm kind of like, well, I will do something if I feel really excited or if it's got that real interest, but otherwise I'm like, no, it's a Saturday or no, it's a Sunday. I'm not doing it.
Noel (25:14)
Yeah.
Katie (25:39)
Okay. And then the last one. So it was about multitasking with AI and automations. Makes you feel productive but actually...
Noel (25:43)
Yeah, I think I do agree with it because like I said, last week I've been in that trap.
Katie (25:44)
Maybe isn't, because of the different contexts and switching and I guess like brain load.
Katie (26:04)
Okay, I think I'm also going to agree and the reason why I agree is because I've seen you do it.
Noel (26:13)
I was doing it just before we hit record on this podcast.
Katie (26:20)
I knew you were. I could see your eyes darting about different screens. Noel sat in front of three different screens, I'd like to add.
Noel (26:24)
Yep. Claude's busy. He's off doing something for me at the minute.
Katie (26:35)
I think what this episode really has highlighted is boundaries. You need to have your own boundaries within your business.
Noel (26:41)
Definitely. And also it comes into education as well. So if you've got a small team that also use AI, it's kind of like training them with that same mindset, isn't it? So they're not going off and doing random stuff. They're creating documents for HR and they're in engineering or something.
Katie (26:59)
Yeah, it's making sure you are looking after your team's welfare if you do have one, but then you need to put that same care and attention into yourself. If it's just you, treat yourself like you would a team. Like, would I expect a team member to be doing this? Would I ask a team member to be doing this on top of everything else they've done today? I think that's really important.
Noel (27:14)
Definitely.
Yeah, but there was one other thing that I noticed within this blog post. It all talks about chat LLMs. It doesn't talk about automations in there. So obviously I put it through Claude and I was just triple checking that I'd read it right. And it was like, yeah, no, this is specifically about chat models. So basically the whole report was looking at about 200 businesses, but the trap they were falling in was the ease of the chat model. So firing up ChatGPT or Claude and asking questions and having that almost instant response. That was the chat time that was taking up a lot of people's time. And it was becoming that way. That was the key issue.
I then started looking at this from a different angle and thinking, well, hold on a minute. They're just using agents because they think they should, because Claude and ChatGPT are essentially an agent. So it's kind of like, well, if those sort of things were done by an automation, you don't have to sit there and watch it. Those sort of things just happen in the background. So you could tick a box in the CRM and then something goes off and happens and then the job's done for you. You're not having to have that conversation and back and forth, which can lead you down different paths as well. With the conversation, you could then think, actually, I could code a tool that could do that. And then before you know it, you're off doing something completely different.
Katie (29:15)
Yeah, I feel like you can go down a massive rabbit hole as well when you're using an LLM because you'll ask it something or ask it to help you with something, it'll spit something else out, and then you're like, oh, I haven't thought about that. Start a new chat. And then you're talking about something completely different to what you were talking about just a couple of minutes ago.
Noel (29:44)
Yeah. So what I would suggest is if you're asking ChatGPT or whatever the same stuff every week, or you have those common tasks, then start to look at putting those into an automation and just let those run on autopilot. It doesn't take that much to do that these days. I would definitely look at that because then you are genuinely saving time because you're not having to type away or talk to it and then you don't get distracted either. The distraction part is quite important.
Katie (30:21)
Yeah, for sure. Easily distracted.
Yeah, I think our attention spans are shorter and shorter, aren't they? And sometimes we're doing things and we don't even realise. For example, we're asking Claude to do something, it comes up with something we hadn't thought about, right, start a new chat. And then you are chatting about something that actually wasn't even relevant. Not on your to-do list.
Noel (31:12)
Yeah. I guess with chat models you've also got the disclaimers at the bottom saying the responses you get back may not be accurate and things like that. Whereas in an automation, you could train it in those specific tasks to give you what you need. That's right. First time, every time. You're not having to go through that loop of editing and what we find with ChatGPT, arguing to get what you need.
Katie (31:27)
Yeah, I mean, I've done arguing now with ChatGPT because we're not friends anymore. Me and Claude are besties now, I'm afraid. Sorry, ChatGPT. But actually, we should mention this. This is one of our kitchen lunchtime chats.
Noel, do you want to tell people what you asked ChatGPT? Was it yesterday?
Noel (32:16)
I think it was over the weekend, actually. But I looked through Substack and someone had posted a picture of a ChatGPT response. I initially read it and I thought, no, come on, that can't be true. Because these days you could put a ChatGPT response into something like NotebookLM and then it could edit it to make it look like a silly response. I think these posts can be easily faked. So I was sceptical.
But the question was this. It was like, I need to go wash my car. The car wash is 100 metres away. Should I walk or should I drive? Now the obvious answer, for a human being that's obvious. I need to wash the car. The car wash is 100 metres away. I've got to take the car to the car wash. That would be the logical thing to do. So I put it into ChatGPT and it came back with the same response I saw in this Substack post, which was, well, walk. I was like, what do you mean walk? And it was going through all these different things saying, well, if you walked, you're saving wear and tear with starting an engine from cold just to drive 100 metres, you're getting exercise. It was like six points that it came back with. I was like, you've completely missed the point here.
So I stopped at that point and sent the exact same message to Claude and it sent me a one sentence response, which was, well, of course you're going to drive because you've got to take the car to the car wash. I was like, thank you. Someone understands it.
But I kept going back to ChatGPT to see how long it was going to take for it to twig what I'm getting at. So I said, well, as soon as I'm walking, should I take a bucket with me? Because I've got something to wash the car. And it was like, yeah, but if it's got the jet washers and the foam brushes, then you could use those. And I was like, well yeah, I could, but my car's still at home. And it was like, yeah, that's a good point. You probably should just take a bucket and fill those up while you're there. I was like, come on. I had about six back and forths before it twigged and went, actually, yeah, no, you should take the car because you need to wash it. I was like, wow.
Use the right LLM for the right task is what I would finish that with.
Katie (34:43)
Yes, that is the lesson we have learned today. Along with boundaries.
Noel (35:04)
Yeah, but now I can see loads of videos of people doing somewhat similar. So they've either asked that same question or they'll ask other similar style questions and then see what they get. Some of the responses are brilliant. Good times.
Katie (35:10)
Love it. Okay, so should we leave it there for this week?
Noel (35:24)
Yeah, absolutely. We've covered a lot this week.
Katie (35:27)
We have. And we hope you've enjoyed this week's podcast. Please feel free to subscribe, give us a like or leave us a review. It really does help us. If you've got any questions for us, please either email us or leave them in the LinkedIn group. We love hearing your questions and we will catch you next time for another episode very soon.
