A simple shift in AI management could be the key to lighter workloads, smoother flow, and finally freeing your time.
You bring AI into your business expecting relief.
Less grind.
More flow.
And yet… the work doesn’t feel any lighter.
You’re still checking everything.
Still reviewing every line.
Still fixing what “might” be wrong.
It looks like progress on the outside,
But inside, it’s the same weight.
That’s exactly what came up in our team’s daily standup this week.
One of my team members has been using AI for more than a year.
On paper, it should have cut her workload in half.
But when we looked closer, the opposite was happening.
Time saved in execution was being swallowed,
By constant tweaking, rebuilding, and second-guessing.
And the realisation hit:
This wasn’t an AI problem.
It was an AI Management problem.
It was the same pattern I often see when people bring on a new hire.
You delegate…
But then you hover.
You check everything.
You jump in at the first wobble.
Not because you don’t trust people.
But because letting go feels risky.
Here’s the part that surprised us:
The same thing happens with AI.
We treat it like a fragile tool instead of a capable team member.
We over-manage it.
And protect against mistakes that may not even matter.
And that “safety” is what keeps the workload heavy.
This week’s episode is about the shift that changed everything for us.
A small mindset change that opened the door to real efficiency gains —
Not by pushing harder, but by getting out of the way.
🎧 Listen now and see if this is the bottleneck you’ve been missing.
KEY TAKEAWAYS: AI Management Shifts That Unlocks Real Efficiency
- Let Go to Move Forward: Trust both your people and your AI. Think of AI as a new team member, train it well and support it by giving it what it needs and retraining until it gets things right. Set the rules, then let it do its thing and watch your free time and results increase.
- Learn as You Launch: Deploy new tools or team members quickly, collect feedback, and keep tuning. You only discover what works by doing – waiting gets you nowhere.
- Fail Fast, Fix Faster: If something or someone isn’t working out, don’t prolong the pain. Make necessary changes based on experience and keep improving your system.
- Let the Numbers Lead: Judge success using real metrics, e.g. email open rates or website traffic, stop being led by your gut feelings. The numbers don´t lie.
BEST MOMENTS: Insights on AI Management Practices
04:12 – 💬 “ When we're creating AI agents to do work, it's just the same as delegating. The same fears, the same reservations, the same micromanagement behaviours can actually get into play, here.”
08:15 – 💬 “The challenge for my marketing assistant, now, is that she's not allowed to review anything that the AI does. She's got to delegate properly and let the AI do its thing.”
12:07 – 💬 “I sacked the old GPT. I literally just deleted it, just like you would, if you sack a member of staff…and created a new GPT to solve that problem.”
14:28 – 💬 “This will be a continuous iterative improvement, going forward.”
TIMESTAMPED OVERVIEW
00.00 Intro: AI Mindset Shift Challenges
01:41 Why AI can become a time suck – the error most businesses make.
05:38 Treat AI like a new employee – delegate, monitor and retrain.
10:15 Checking results and tweaking.
Episode Transcript
Please note: This transcript was generated using automated transcription tools and may contain typographical errors or inaccurately captured words or phrases.
Dr Steve Day: AI is transforming the way we are working in our business. But it\'s not all been plain sailing. We\'ve had many challenges along the way. And many areas where we\'ve tried to use AI to improve the efficiency and the effectiveness of processes. It\'s actually ended up taking far longer than we expected. And in some cases, is to mean that the task is taking longer than it did before.
We had a huge mindset shift or a huge realization in our daily standard today. That made me see, one of the biggest challenges, I wasn\'t aware of. A blind spot, both myself, but also for my staff. That what was preventing us from getting the true value we can from AI. Today, I want to share what that mindset shift was. and how it may be the bottleneck. That\'s stopping you from getting the full benefits as quickly as you would like from adopting AI by both yourself and your team.
And if you like this episode, please remember to hit subscribe. So you don\'t miss future episodes on how to live with more presence, more purpose and more freedom. And to make your business run more smoothly with or without you.
Okay, so in our daily stamp today, we were talking about an AI GPT that we\'re creating. And this particular one was about doing the titles for this particular podcast. This is something that we have worked on for a long time, along with how we write our emails. How we write our show notes. How we edit our transcripts. All this kind of things.
We\'re now adopting AI to massively improve the process well. That was the intention anyway. But when I recently reviewed the time sheet of the staff member who is actually responsible for doing all of that. So basically she does all the copywriting for the entire podcast production and the promotion. And she\'s been now using AI well for well over a year. With intention in my mind, to massively reduce the time taken to actually produce all of this content.
The reality has been different. The reality has been that it\'s not really affected the time. Because she spent a lot of time building these GPTs. Of testing them, of tweaking them, of improving them. And just continually working on them. So all the time that may have been saved on actually. Doing the work in itself.
So she was just physically like, or manually like creating this copy or whatever. That\'s actually all been eaten up by the development side of the AI. And I\'d hope by now it would\'ve improved. But it actually hasn\'t. And today we had a big realization about what I believe. One of the hugest bottlenecks in the gaining or experiencing the benefits of applying AI to a process.
That I was completely blind to. I realized in this conversation with my, marketing system, that I\'ve actually been guilty of this as well. And let me explain exactly what I\'m talking about. So when my marketing assistant has been looking at a, we using AI, she is the, just to put it in context, she\'s the kind of person who is incredibly conscientious.
She really cares about her work. She spends time in reviewing things. She\'s got that incredible attention to detail, but she also means that she tends, takes. Time over doing her work now I\'m all for that. I\'d much rather this work being done well and properly than just rushed out and it been a complete mess.
However, there\'s got to be a balance. And in this particular case, what was happening was a the same. Issue or the same challenge. I see so often when delegating to staff and it, the realization I had, the mindset shift I have was when we\'re creating AI agents to do work. It\'s just the same as delegating. And the same fears, the same reservations, the same micromanagement behaviors can actually get into play here and mean we don\'t get the benefit.
So just like when you hire a new employee. And you haven\'t built up a lot of trust with them, you don\'t know how good they are. It can be hard just to delegate work and like hope for the best. You know, you can do your best and try and delegate well. And you can try and give good instructions. But ultimately this is someone you don\'t know well.
You haven\'t, got that, you know, that level of trust, that you know, they\'re going to basically do their best and fix problems if they occur, et cetera. So you might be inclined to start micromanaging. Looking over their shoulder, checking in all the time, reviewing all of their work. So when you actually hire this person to delegate work to, free your time, what you end up doing, is actually doubling your work.
Not only have you got to delegate the work and you\'re managing a person. You\'re now reviewing all the work they\'re doing as well. It can end up taking more time than it would\'ve just taken to do it yourself. And you can start feeling like it really isn\'t, wasn\'t worth the effort.
The same issue I\'ve seen with myself, and in this case with my market system is happening with ai. The problem is that my marketing system doesn\'t trust the ai. So what I found was when I spoke to her in detail about like, why is it taking so long still to do this stuff, I thought we\'d, you know, built these GPTs.
I thought they were working okay. And the problem was that she was, yeah, using the GPT to help her do the work. And then she was basically doing all the same work she\'d done previously. To make sure that she agreed with what the GPT had created now. In the development stage, like with a training stage of a brand new employee. There\'s an argument to say, yeah, maybe, you know, checking some things. And making sure it\'s, generally aligned. And making sure the standards are okay.
Of course, that\'s going to be important. We don\'t want to just sort of, abdicate responsibility and hope for the best. We want to be monitoring, like, what is the output that\'s personal in this case that this AI is producing. If we run the risk that if we don\'t let go. If we don\'t actually trust the thing that we\'ve created or the person that we\'re delegating to, then we end up in that trap of actually taking more of our time rather than the, the objective. Which is to free our time.
And that is exactly what happened here. So the. Conversation we had today was about, okay, well what would happen if you just didn\'t do any checking? If you just trusted exactly what it produced without going in and looking like, in this case, like listening to the podcast itself. Understanding like what the topics were about, the main points, the things, the salient stuff that we don\'t want to miss. Like making sure that the AI is picked up on all these nuances, all the stuff that could so easily be missed.
Like I said to her, well, what would happen if you just didn\'t do any of that? If you just trusted like what it actually came out with. And this is exactly the same psychology. Or mindset that\'s needed when you are delegating to somebody else. Is that you\'ve just got to let them get on with it. You\'ve got to let them make their own mistakes and then.
Pick up on those mistakes, give feedback, and then coach them on improving them if they can\'t fix it themselves. And that\'s the approach that I have been taking recently with AI. And seeing massive leaps in how quickly I\'m getting stuff out there, getting it tested, seeing the result, and then going into it.
Iterative, you know, in process improvement exercise and giving it feedback just like I would a human. So that was the advice that I gave to my marketing assistant today, is just go with it. Like monitor the metrics, see what the outcome is like, whatever it is we\'re doing. Whether it\'s creating copy for marketing.
Then we can monitor the metrics. We can see does it affect our open rates? Does it affect our click-through rates? Does it affect our reach or our traffic to our site? We\'ve got like real data we can actually monitor to say like, is the AI doing a good job? Like we, our opinion may be wrong.
And the same way with we\'re hiring an employee. Our opinion about the work that\'s produced may actually be wrong. It could be just subjective and actually the results that you\'re going to get or the results you get as a result. Of whatever work is produced, that\'s what\'s important. So laying go of this need to actually have your finger in every pie of actually micromanaging everything, of being, you know, the one that makes a final decision.
Everything can be hugely liberating for us as managers and business owners, and you never know we might end up with work being done. Getting a better result than if we did get involved. And waste our time tweaking and basically, yeah, just basically meddling with something that didn\'t actually need to be fixed.
So the challenge for my marketing assistant that I set her this week is that she\'s not allowed to review anything that the AI does. She simply just got to trust what she\'s already set up. She\'s already tested it. She\'s been through, she\'s, you know, seen some work in the past that it\'s done. And it\'s been okay, whatever.
She\'s just now got to let go. She\'s got to delegate properly and let the AI do its thing. Then we\'ll then see what the results are. If the results show that this is a catastrophe and you know, our, open rates via emails just like plummet through the floor or the traffic on off site goes down or download, suddenly drop off, of course we\'re going to jump in and actually fix it just like you would with a human.
But if it\'s just small little errors, things that pick up that actually we think, well actually, you know, does it really matter? Then maybe we\'ll just let it run or maybe we\'ll give feedback. Just like we would to a human. We\'ll explain what we expected. We explain the issue we\'ve had, and we\'ll ask the AI to actually improve itself just like you would a human.
And this, treating the AI as a human in this context, it sounds a bit weird, but actually from my experience over the past few months of adopting this behavior with how I\'m developing my GPTs of all different areas of our business, this is what I\'m seeing of getting really big results. It\'s.
Doing the initial work setting up just like you would. When you bring someone new in, you train them up, you give them their basic training, you delegate the task, you know, you give them their training videos, their operation lines, whatever they need. And then you let them get on with it and see if they sink or swim.
And if they swim brilliant, then you\'re onto a winner. If they sink, you can give \'em support and coaching. And if they\'re no good, you basically let them go just like you do with an AI agent. Like this morning I woke up super early \'cause I was a bit excited about this \'cause I\'m a bit of a geek. And so we\'ve got this, we\'ve created an AI agent, you know, I\'ve created an AI agent that, sorry, not an agent, a GPT. A GPT, that basically writes all of our operational manuals for us.
So we have a process called DIDACT, which I\'ve mentioned many times in our podcast. In the podcast, sorry. And it\'s a method of framework. For creating super efficient operational manuals. And we also have a framework for checklists as well, for making effective checklists. And what I\'ve done is to train this AI, GPT to take a raw video of someone doing a task and arraying it through, stick the transcript in, and it produces this.
Totally standardized, formatted, what we call the DIDACT operation Manual, which captures everything needed to both delegate and do the task first time and every time Now. It\'s been okay, it\'s been getting okay results for us. It\'s a bit clunky sometimes, get things wrong. We got some feedback from clients that have been using it, same issues that we\'re facing, but what I didn\'t do was spend weeks and weeks developing it until I had it out into the world.
And so we started using it. I actually created something, you know, relatively quickly. Got it out there, and then a few weeks later we are then reviewing the results of that work. So I\'m looking at the metrics that I\'ve got, and in this case, it\'s my own experience, my stats experience, and my client\'s experience.
I then woke up this morning, as I said, had an idea and basically, like I have said, I sacked the old age, the old GPT I literally just deleted it. Just like you would if you sack a a, member of staff, a member staff is not working out after the first few weeks, then, you know, despite you actually supporting them, giving them coaching, that\'s the time to let go.
Actually, I talked about this in the previous episode. When I actually hired somebody that wasn\'t suitable for the role. And I decided to let them go, and actually it felt good to let them go. Because it left, didn\'t leave any bitterness in me that I had the wrong person. I was supporting them too far, and actually, it, was in, it was inspiring to see their reaction.
Which was actually, they were motivated to go away and learn more. To be able to actually come back, hopefully, and do the role in the future. So we\'ll see if that happens. But that would be a brilliant outcome in this case. Anyway, with the AI, with the GPT that I\'ve created, I basically sack the old one and said, you are no good.
You like, you\'ve given you like enough time and effort, so we\'re going to hire somebody new. In this case, it was actually just create a new. GPT to solve this problem with all of the learning I\'ve got from that one not working. Of all the mistakes I made with how I set it up. Just like when you hire somebody that\'s wrong, you learn what was wrong about them.
You then, when your next half of the role, you\'ve got a better understanding of who you need in the role. I had a better understanding now of how to develop this GPT. So within a couple of hours, actually probably about four hours, in fact, messing around and getting it working, I\'d actually created from scratch, a new GPT, far simpler, far more focused, and it did the job way better and quicker than the old version.
And, so far it works really, really well. Again, I\'ve not gone and spent hours and hours testing and retesting and monitoring it and like reviewing everything I\'ve gone. Yep, that\'s good enough. I believe it works. So like how new person I believe they\'re going to get on with it. for, that analogy, I\'ve then put it out to the world.
I\'ve shared it with our clients, I\'ve shared it with my staff. We\'re then going to like monitor how it performs. Look, I\'m going to give it feedback just like I did in the testing stages when I was just initially building it. I gave it some test data, saw the results. And gave it feedback until it actually tweaked to it actually gets pretty consistent, good results.
So we\'re going to go forward. We\'re going to continue doing that. I\'m going to be monitoring the, results, the metrics that we get from it. I\'m going to then be feeding that back in if there\'s any tweaks that are going to be needed. And this will be a continual iterative improvement going forward.
Doing it this way, getting stuff out there and actually monitoring and actually tweaking as we go forward is, I believe a way of they\'re not getting involved with. As I say, like micromanaging the agents or the ChatGPTs. Just in the same way as you shouldn\'t micromanage your staff, you\'ve got to let them single soon. They\'re either going to get it right or they\'re not. The only difference here is we\'re more in control of the, initial setup of the thing, but once it\'s there.
We\'ve got to just let it go in my opinion. And let it actually figure stuff out with the feedback that we can give it continually to see if it can improve. So that\'s it. That was my massive mindset shift that I believe is it will hugely reduce the amount of time taken for our staff to start adopting AI this, understanding that we\'ve got to treat it like a new employee.
We\'ve got to give it the support in the beginning. We\'ve got to give it the training in the beginning and then we\'ve got to let it go. See the actual results, the metrics that what we can measure to see if this is actually doing the job as good or better than we did previously. And if not, then we can do some feedback or whatever or improve it.
And if it still completely fails and we like, it\'s just a disaster, of course we can sack it and start again, or not hire into that position or not build a new AI if we think actually this is an inappropriate use of ai. Or you know, same way for an inappropriate hire, then we don\'t actually waste our time doing it \'cause we learn the lessons.
Actually, okay, maybe this wasn\'t the best use for of our time right now because the cost versus benefits, you know, aren\'t really there. So if you found this useful. Please do comment. If you are watching on social media. Let me know your thoughts. How are you adopting AI? What challenges have you faced and how have you overcome them?
If you have found this useful, please remember to hit subscribe so you don\'t miss out on future episodes. And share this with anyone you know that you think would like to learn how to do more wonderful things. To make yourself live with more purpose, presence, and freedom in the world. And create a business that runs with or without you.
Take care. Thanks very much for listening. Bye-bye.
LINKS TO CONNECT WITH THE HOST
- Podcast: https://www.systemizeyoursuccess.com
- Website: https://systemsandoutsourcing.com/
- Facebook Group: https://facebook.com/groups/systemsandoutsourcing/
- LinkedIn: https://linkedin.com/company/systemsandoutsourcing/
- Instagram: https://instagram.com/systems_and_outsourcing/
- YouTube: https://youtube.com/@drsteveday42
- TikTok: https://www.tiktok.com/@drsteveday42
ABOUT THE HOST
Steve moved to Sweden in 2015 and transformed how he ran his businesses—switching to a fully remote model. A former NHS doctor, with a background in computing and property investing, he now helps overwhelmed business owners systemise and outsource effectively. Through his courses and coaching, Steve teaches how to automate operations and work with affordable virtual assistants, freeing up time and increasing profits. He runs his UK-based businesses remotely with support from a team of UK and Filipino VAs, and is passionate about helping others build scalable, stress-free companies using smart systems and virtual support.
For more articles related to AI management in business, you may also like:
How to Keep Your Team Relevant and Future-Ready in the Age of AI
How We Leverage AI Apps and Virtual Assistants to Put Our Content Creation on Autopilot