Personal Automation with AI

Mark Smith (00:11)
Hello and welcome back. It is Monday, October the 20th for us doing this recording. Remember, we're in the Southern Hemisphere, so we're slightly ahead, but this is the AI Advantage. And this week we've got so much community feedback in the WhatsApp group. It is amazing. And so we're going to share a bunch of those stories, but our focus today is very much on how do you personally use AI to add, is it a layer of automation into our lives? Right.

Meg Smith (00:41)
Yeah, that we've talked about personal automation being something that AI can help us enable. But maybe before we jump into that, it's worth getting your hot take on automation in general. Cause I've got my take where it feels like everyone is screaming about like, I've got this agent, I've got an army of agents to work for me. Everything in my life is automated. And to me, it sounds like a bit of vaporware, a bit of like over-hyping certain things. So today. I'm really excited to look at these examples of how people are using AI to help them practically in their lives, I suppose. But what do you think about the commentary around automation?

Mark Smith (01:18)
I like it. Yeah, very, very cool. Very cool. The, I think that right now we're in agent slop land where everyone's got their agent tools. And when I look at any of the agent tools in the market where it doesn't matter, you name the company that's providing them. seems that they require me to do a whole bunch of work to save myself a whole bunch of time. And I'm like, hang on a second for me, that just doesn't resonate. The whole idea is it should save me time and that's the time to set it up as well. And so I think that we're still in the hype phase of automation and agents and what agents can do for us. Yes, there's some edge use cases that are proving it in a commercial sense, but once again there's so much work to be done before you get a usable, repeatable output. And that is my, suppose, right now I'm like, there's a lot more work that needs to be done in agent land to get to that point of what I think true agent or a Gentic based support from AI, you know, driving real value. And let me give you how I see it.

If I want an agent built, I would hope that through a series of conversations with an AI, that it would build me, actually not just conversations, but observations. It would be able to build me an AI that was robust, secure and met my needs. And I think RPA, robotic process automation, is part of making that work. And what part is it? It's the record function.

For AI to observe what I'm doing on my computer, for example, let's use the desktop experience and sees me, let's say it recorded the entire day of what I did. And then at the end of the day, it said, hey, listen, I noticed that you do a bunch of this stuff. In fact, if I look back over time, you repeat this over and over again. And I've created a way to help speed this up for you. Not me going and telling it to, but it's created a way that says, hey, let me run it past you.

Let me give you a demo. Let me give you a show and tell you can tell me where you'd like it tweaked, et cetera. And game on and we iteratively build out based on the real stuff I do every day, how I handle email, how I handle scheduling, how I handle product. Productive work, right? When I'm really in my, you know, my focus time blocks of a day, how could it improve? And I'm wanting it to be this, you know, alongside helper that's always looking at how it can make my life more friction free from the drudge work. And that's what I'm looking for from an automation tool.

Meg Smith (03:57)
It reminds me of it's Meredith Whitaker, who's the CEO of Signal. And she talks about agents and the way that people are talking about them and using them and kind of highlights that in order for agents to do exactly what you, what you described, that requires you to give over a lot of control and insight into your personal world. She's saying, like thinking about like,

being able to spend money on your behalf if you provide credit card details and things like that. And I love hearing that balance or that thought of how we think about trust and how we think about what systems do we trust with this information? And at the same time as going, let's be experimenting so that we can be moving forward and getting the benefits. There's a really great...

you talked about edge case, right? Specific cases. There's one in New Zealand called Pam. The ⁓ founder is Nicole Ritter and she is a Māori CEO and she got sick of carrying all the mental load of family admin, of school information, of meal planning, of all these things. And so she made a personal assistant for that work, right? And so in those scenarios, I think...

you know, there's a lot of people are going great, I need that. And actually one of the pieces of feedback we got from our community from James, think it was, who was saying that he was helping his brother use Gemini, Google's Gemini to make sense of communications that come out of school that kind of, just, his brother needs to know the so what, what's the so what from all of this. So yeah, I think that what I,

We also had a lovely comment from someone say ⁓ that listens to the podcast that said they love how you and I play two sides of the coin. Like obviously I'm more, my risk appetite is more conservative with these things, whereas you're very keen to play an experiment. And so I think that's kind of a, a, ⁓ a note for our listeners as well, for you to think about your own risk appetite. And as long as you're, you're conscious and aware of the choices you're making.

I think it's, it's sort of, if you're happy to take the risk, go for it and then see what your experiments, you know, I don't know, result in, deliver.

Mark Smith (06:08)
Yeah,yeah, that's fair, that's fair. I I love the pace of change and the opportunity to continue to learn. I mean, I woke up this morning and decided I need to, and I didn't just wake up and decide this morning, but I need to lose some weight. And so for me, my effective way of losing weight is to do fasting. so,

some of my more recent research and by the way, I've been fasting for years I've done two 21 day water only fasts in a single year. I've done multiple seven day fasts of water only and Then I've done Incremental fasting for multiple years so I and it's a great way for me to get into ketosis and to control my My

blood sugar spikes that I get and of course it has a real positive effect. Anyhow, I was just like, I opened up an app that I've used for ages, it was called Zero. And when I went to set my custom routine for this time, it just kept crashing and crashing and crashing. this is, right now it's 9 a.m. in the morning or a bit after and we've dropped our kids off at daycare, but this morning I built an app that allows me to carry out my...

Fasting is in what I you know, basically tailored to me. And so here you can see it on screen It's called fast flow, which the AI made that up ⁓ It's got a tracker in it. It shows us color palette, you know, I can pause the fast I can resume the fast There's you know, my next eating window. This is some sample data in here. It's got a schedule

with a kind of where I'm up to in my particular timeline now and my windows and my second increments for that. I can see a progress update, how I'm doing fasting. I haven't tied my weight into it here yet, but kind of gives you a look feel. And then it's got a feed coming in here of tips to help me stay motivated and have the right mindset, et cetera, while fasting, right? Pretty straightforward, heaps of tweaks and work needs to be done on.

that ⁓ app. But it shows how quickly things are changing. By the way, I built that on a new AI tool that I was only exposed to yesterday called Kimi. And Kimi is out of China and it's phenomenally amazing how quickly I was able to give that a prompt. It spun up a computer and did nine different phases of work to actually build out that solution. Now I could go back and prompt

changes, et cetera, into that experience. But that's as far as I got before dropping the kids off. So I feel that we're going into this phase of, as part of, know, for me, why is that personal productivity? Because I'm a numbers person. I like to see data. I like to see it visualized and I like to track how I am going. But I don't need all the marketing and crap from the other apps that are out there. And so,

How quickly was it for me to build an app just for me? It doesn't need to be for anyone else. don't need to make it device independent. not turning it into a business. It's just for me. It's to achieve a goal that I want to do around my health. And I love the beauty and the speed that it is possible. I didn't have to tell it what JavaScript to use or whether to use Next.js or whatever It just shows the best to drive the outcome that I needed to produce at that point.

Meg Smith (09:36)
And it put me in mind of that. There's a clip of Eric Schmidt, who was the, ⁓ the CEO that Google brought in, after Larry Page and Sergey Brin had kind of taken the company sort of so far. And then he was the guy that they brought in as CEO. And I think he was chair of the board for a while after that. ⁓ he was being interviewed, and said that AI is going to allow people to build the Tik TOK themselves, the app, if they want to.

in a few sentences. And I remember seeing that and it felt far off. It felt like, yeah, okay, I can see that being a future. And I was particularly interested in the like, but why would you build your own TikTok app? And what's the point of going to a party if no one else is there kind of thing. Like that was where my mind went. But this example this morning, and I saw your prompt, it was like two or three sentences. And then I also have,

observed your fasting journey and seen the different apps you've used over the years and how, you know, you kind of have been limited by the Apple store, the app store, whatsoever is in there. And you're always kind of jimmying an existing app that someone else has designed that doesn't quite fit your particular goals or your particular way of doing it. So that was a bit of a moment for me where I was like, this is what he meant. Like that the way that we can design.

software that is specifically to help our own personal productivity or our own personal purpose, like to serve the purpose that we're trying to go after. Cause to you it's, it's more, I think we have to keep coming back to the like the why we get so caught up in talking about the what like, great. I can build an app. But why are we really clear on knowing the purpose of that? and if you can start in a, in a personal context,

I think the learning we have there will mean we're so much better at doing that in our work as well.

Mark Smith (11:25)
Yeah, yeah. We also got a bunch of folks that sent how they're doing personal productivity through this week. And so I thought we'd start with the first one here from Guy Orange. He said, I've used AI a couple of times recently to help with scheduling or on training. I'm attending an online finance course called Rebel Finance School, Wellworth Look, which is 12 weeks plus exercise homework. So I uploaded the content.

timetable and had it review my availability and my question on suggestion how much time to dedicate per week and when I would like to finish the course. As I have a rule to challenge my assumptions in the model, it pushed back on how much time I should invest. Once I was happy with the outline, it created ICS files for three to four sessions a week, split over week, night.

and the weekend along with a study checklist and a summary document with links to all the videos and worksheets. Now what a brilliant way of using it to drive that productivity but not destroy everything else you've got on your schedule and your life and things like that and putting that balance ⁓ in place.

Meg Smith (12:33)
It's so good. I've been writing a series for our newsletter, Tech Trailblazers. I've been writing on learning and this to me is just such a great application of this idea of using AI to help you learn. And sometimes when we talk about that, we mean, you know, to help you learn a particular topic of interest to a task that you're trying to do. This is so much bigger than that. It's about how do you fit in a certification or a training program?

or a uni paper or whatever it might be that you're going after. How do you ⁓ plan and make the time to do that? I love particularly the creation of the ICS files so you can add it to your calendar and those ICS files containing the relevant worksheets. Super cool. I also did go up and look at Rebel Finance School. How epic. So it's a UK couple who achieved financial independence in 2019.

Mark Smith (13:18)
Yeah.

Meg Smith (13:26)
So they've created an offer this 12 week program for free. So it's not like a, you know, buy my financial literacy course where I never talk about the fact that I made my money by selling financial literacy courses. It's a genuinely, we've learned this, we've benefited from this knowledge and now we're going to give it away because it can benefit so many other people. So there were a couple of other people, Emma Claire mentioned that she's doing it as well. Yeah, so that was a good shout out. As a specific example of how you can use it for training and learning support.

Mark Smith (13:58)
I that. On the back of reading guys, you know, submission there, I had a different AI this morning pop out my ICS files with my schedule and the calendar for fasting because I'm doing this fast that is 36 hours fasting, 12 hours eating window, then 36, so it rolls through the week and I optimized it to.

as much time as in my sleep zone. So it starts at 9 p.m. at night and it finishes at 9 a.m. in the morning 36 hours later to try and get my eating windows in the daylight hours and my sleeping windows maximizing those fasting windows. Just why don't we make it as easy as possible? And so I exported the ICS files just like he suggested and what he did and imported them into my calendar. Another great one that we had came in this scenario for a one-off

purchase something that you wouldn't necessarily do very often. It came from James Sloan. It says, I summarized searching for camper vans. He's based in the UK with OpenAI. Five Minutes has set up a process that scrapes all the different vendor sites and looks for specific van configurations we were after. ⁓ The ⁓ mechanic side of it, the more vague terms like U-shaped seating. It then fired the update list over to me once a day.

found the one we brought from a site that he had never heard of before. And so really going outside of, where your bias or your tendency to go with what you know to find something that was totally unique. Once again, a really great use case of personal productivity for a one-off purchase.

Meg Smith (15:34)
think that this ⁓ specific phrasing like the U-shaped seating, the jargon, the category specific knowledge is such a great use case for AI generally, because it's almost like any time you did this, you would have to learn a whole new language or a whole new vocabulary, and you wouldn't even necessarily be aware of what those things are. So starting with using its broad knowledge to extend your knowledge on a particular subject, I love.

And then I love the aggregation element of it, because it reminded me of when I was working at Google, one of the clients I worked with was a company called Online Republic. They're a Kiwi company that were in the business of aggregating information on the internet. So they had, you know, the sites that were personalized, sorry, that were localized to different geographies that would aggregate motor home hires. So across, if you want to go and drive from, you know,

Seattle through to California, which is a really popular tourist drive. You could go to their site and they would aggregate sort of across all the different bespoke motor home companies so that you could get the best deal. That's essentially what James has asked open AI to do in this purchasing cycle. Right. But it was wild to me because that, ⁓ business online Republic, they.

created so much value in being that aggregator that they were actually bought for $85 million by Webjet. So this idea that we've had these software companies, these web experiences that have, you know, over the years created so much value by bringing that information together in a new way, that now the next step is we are able to be really custom and specific to me and exactly what I'm looking for in the space of a few sentences is just...

kind of mind blowing. But I also thought again, coming, putting on my like cyber security, personal security information. I love James's example because a camper van purchase for me, am I even at my sort of conservative risk of the information I share? I'm looking at that going, I don't really mind if open AI or whatever other service you might be using knows that I'm looking for a camper van because it's a kind of one-off purchase that I can then evaluate.

The benefit to me of getting access to that aggregated information far exceeds any sort of risk of sharing that I'm going to be a camper van owner and that I'm open to buying a camper van. I kind of put it into a bucket of say, Hey, I probably wouldn't do a similar process at my personal risk level for say health insurance, because the information that I would have to put in about say, you know,

I'm actually quite happy to share my information, but if I was responsible for doing it for my whole family, and actually I had this experience speaking with a broker who was using Zoom to record our meeting, and he started going through the questionnaire form with me for health insurance and asking about medical history of me, my children, Mark, which I can say, you I don't mind sharing his information either, but particularly for our children, I...

I just actually stopped him. said, Hey dude, I know Zoom's terms of use. I don't want to be recorded by Zoom telling you all this information about my children. So I probably wouldn't do it for that particular purchase myself, unless I was really confident that the service I was using was going to protect my privacy. But Cambervan purchase, such a great example.

Mark Smith (18:49)
Yeah. The last one I want to share is this one from Simon Doye and he said, have a daily copilot schedule prompt, gives me the top 10 news stories, top 10 business stories from the US, UK and the top 20 tech stories. Sorry, top 10. TV and movies and finally the weather. In the prompt I explain where I'm based and the kind of news and resources I'd like. This was added ⁓ as I was getting quite random news feeds when I didn't have it.

This is helping me cut down on the amount of time I spend keeping on top of things. Once again, an absolute superb example of personal productivity for something that probably happens ⁓ over and over again. I know for myself, I'm very big on staying current, right? I wanna know what's going on in the world. And so when I meet with people,

I can, you know, it's podcasting or whatever, I am current on world events, but I also go back to who's making me want to believe this, right? So, and who's making me want to believe it became a bigger and bigger thing that I noticed was happening in my life, particularly through social feeds is often there are these, you know, intention grabbing headlines, but they were not.

entering other news cycles. And so I like what Simon's done here. He needs that information. And it's something I've done myself. Here I have got a pro version of perplexity. And it says your daily top 20 global AI news. Now, if I go down here and have a look, you will see, see why haven't I got any notifications yet? Let me just refresh. Oh, there they are. So you can see I've got a range of scheduled

⁓ searches that happen for me every 24 hours. here's the list. So four hours ago, it runs at 5 a.m. my time. I have one that says, you know, it's consulting and big tech AI research. there any new white papers? And in the audience list I have behind this McKenzie. It is all the big players, IBM, Accenture.

PWC, et cetera, that have big research divisions that are feeding into this, things like Gartner, Then I've got one here, my daily intelligence age AI brief. So the intelligence age AI brief, if I had jump in here, now we're just coming off a weekend. So I've got to tell that if there's nothing in the last 24 hours, just say nothing, right? Just don't have me wander through rubbish.

make sure and I've got a whole bunch if I go into you know the settings of any of these you'll see here my prompt, ⁓ how I specify it, the frequency, the mapping, the kind of headings that I wanted it to be put under and then what to include and the format and things like that. And so I have it for my world news feed but I also have it for what I call non-western news feed because I feel that once again our western news is particularly skewed to

the propaganda of the West. And so what I wanted to do is stop just listening to, you know, what we're fed in the West. And so now I get a Russian newsfeed, I get a China newsfeed, and then all the other smaller across Arabic nations and things like that in my newsfeed, because I don't want to be sold the line of what people want me to believe. I want to hear more broadly.

And some of the, you know, ones that there are no one propaganda and I find it's funny in the West we go, Hey, this news agency is a propaganda agency for Russia, for example. And yet we don't call out the CNN Fox News and all the other ones that are there perhaps propaganda agencies for other government interests. so what I want to do is, and I have it Mark, if it's known as a, it's coming from a

no one propaganda source, but then is it being triangulated in its reference and it actually puts it to me in the prompt. The other one is particularly around flood, earthquake, fire, risk, all those types of things. tells me it is, it marks it up as the impact it could have on me personally. So I've got a whole bunch of rules around how could this impact me personally. So it does some analysis on it as well. So Simon, I love that, you know, that you're doing a similar type thing there with how you're getting your daily news feed.

Meg Smith (23:07)
I love that you guys are focused on news and I do think it's important and I do agree with the things you're saying about source triangulation and critical thinking and having those personal checks that if you're reading something and it's making you suddenly feel like you've got a real strong feeling because that's what media is designed to do now. You should have a really strong feeling towards one group of people or one type of person. We can check ourselves and remind ourselves that we're, you know, we have more that in common than we have that divides us.

But my thing I took away from that was I should be running a scheduled prompt for TV and movies based on my preferences because by the time I sit down at the end of the day and I want to decompress, we did this last night. This is, you know, we're like, we should watch something. We've been away. We are home. We've done, you know, the, the, and we've packed things. We've done washing. We've put tired kids to bed.

And we're tired and like, should watch something that is like, you know, maybe we, feel like my genre of movie at the moment is like Happy Gilmore is the sweet spot. Like for both of us, we like it, but you watch it once and like, what are you going to watch next time? And of course we didn't, we just sat on our phones, you know, consuming our own personal media because we didn't know what we, we, the choice of what movie or TV we were going to watch that would hit that kind of decompression goal.

was too hard. now I'm looking at Simon's tip going, okay, I'm going to set a TV and movies feeding at my preferences that I know that are things you and I would both like. So that when we get to the end of the day, I've already got something lined up. I'm doing my own personal scheduling. I'm going to call it a TV guide because that's what the little book is to be called that my Nana would choose her TV line up from.

Mark Smith (24:39)
I like that.One of the automation apps I do mainly the grocery shopping in our household. And I know that between these two major brands that represent any sub brands in Australia, sorry, in New Zealand, I forget I'm living in Australia now. And what happens is that they alternate week in week out their discount. So they'll discount and it's almost like they've, although it's, you they wouldn't say they have a monopoly, they have this unspoken thing of, you know, if butter is,

Meg Smith (25:07)
They have a

Mark Smith (25:10)
X price this week, it'll be discounted at the other supplier and because everything's delivered and you know, we've got accounts with those two major suppliers, I want to do an app that says here's the things that we consume in our household. It's this brand of butter. It's this brand of momite. It's this brand of peanut butter. It is these type of sausages, blah, blah, blah, blah. I want to create an app that automatically taking into account

the delivery fees, the discounts, et cetera, that will aggregate in just for me to my personal home preferences of shopping and always optimize my buying experience. I'm not looking for choice people. I'm not trying to be on a Monday night looking for choice and trying my head at a new ragu or something like that. I just want what we eat as staples in our household. And so no creativity, just help me.

get rid of this thing that happens every week and if I forget to order, auto order if I configure it to allow it to do that on my behalf.

Meg Smith (26:12)
There was, this is the problem of living in a country where we have a supermarket duopoly. There are only two, they own all the brands. but there was a really cool guy and his name is Miguel Barkley and he is in the UK and he had an Instagram account called one pound meals. And he would use the app that he built, which was a comparison app across all the different supermarkets in the UK to find the best price, you know, produce in and items. so that you could create a really affordable, healthy meal for one pound to serve. And I used to be so obsessed and jealous watching that going, I it doesn't really work here, but I love the idea of it.

I loved all the feedback from the community this week. It is so fun in the WhatsApp group to see those use cases coming through and other people learning. Like that's the best thing, right? Like we keep coming back to this concept of Aqours, learning and teaching, learning and teaching. We're always learning from each other as we share what we know and what we've learned. So if you want to come along and join us there and share your ideas, it's going to be...

Mark Smith (26:54)
Yes.

Meg Smith (27:14)
Really interesting next week as we look at, for the first time talking about collaboration and delegation. So this is the human AI teams that are starting to come out more and more in some of those reports you were talking about, Mark, from the likes of Microsoft and Gartner and PwC.

Mark Smith (27:32)
I like it. The WhatsApp link if you want to join the WhatsApp group is in the show notes. Remember if you are listening to this as an audio only podcast it's available on video on Spotify and YouTube as well. And with that we will see you next week.