Intros and AI Hype

you

This is Unpivot. Welcome to episode one. I'm Wyn Hopkins. I'm Mark Proctor. I'm Sue Bayes. And I'm Jars Mayall. So the concept of this show is that we're going to be discussing all things Excel, Power BI and other business related topics.

uh it's really just an excuse for us to chat about things that we're we're nerding out about shared passion about data and business value we're gonna focus on business users rather than i.t but everybody's welcome it's going to be casual we just want to have a bit of fun we'll have future guests dropping in

and we'll be doing future live streams at some point so you the kind listener can also take part all right so given this is episode one i think we should just do a brief intro of ourselves so mark

Who are you? Where are you? What do you do? I'm Mark Proctor. I live in the UK in Peterborough, which is about an hour north of London. Having qualified as an accountant many years ago, I now... try and help accountants to automate their month ends get their work done a lot faster get their budgets and forecasts done faster so they can save time and go home on time perfect sue

I'm Sue Bayes and I am primarily a Power BI person, although I also am invited in because I love Excel. So I run my own business, been doing so... a few years and people pay me to help them with their data solutions and I'm a very, very keen runner as well. So you'll hear me chat about that a lot. But I just like having fun, really. And data is fun. So that's why I'm here. Awesome. And Giles.

hello uh i am co-founder of full stack modeler which is a training company for excel people for finance people financial modelers i'm 10 years into my midlife crisis, which means I do silly things like ultra marathons. And I'm increasingly annoyed by how good Sue is at running when she's only been doing it for a couple of years. 20 years, actually.

Oh, okay. Oh, you lied to me. I thought you were just new into it last year. Running for 20 years, but the last year or so, I've only run marathons. Faster than me every time. Yeah, faster than you every time. I'm not breaking. that's fine it's the phrase i in the last year i i only run marathons

How quickly can I talk about Comrades, Wyn? I haven't showed off my medal yet. Wyn hasn't said who he is yet. I've got to introduce myself. Can we at least get to the formalities until this turns into a running podcast, okay? Sponsored by RunKeeper. Any other name drops for a few bits of sponsorship? My name is Wyn Hopkins.

director at a 10-person consulting firm called Access Analytic in Perth, Western Australia. I'm an ex-accountant, but now I specialize in Power BI, Excel development and training. The company also sort of delivers a budgeting system called Solver. I've written the book Power BI for the Excel Analyst and have a bit of a YouTube channel that's quite popular. So search the Access Analytic channel. Okay.

it's a little bit popular a little bit popular well it's relative isn't it it's not mr beast okay yeah um so um before we get into probably more topics about running rather than anything else but stick with us all you non-running fans me and mark will keep you on the straight and narrow we will yeah so uh quick question random question uh i'll aim it at

sue to start with what was your first job out of school oh um rank xerox i worked for rank xerox if you want my proper job i mean i used to make candy floss and ice cream i live in cornwall so i used to do like candy floss and ice cream and stuff but my first proper proper job was working for rank xerox and because i'm very very old um although i know i don't know kids but

But it was when I first experienced the windows and being able to use an intranet because I was based at Wellingodden City and we had a site down in...

Oh, Mitchell Dean. And so I was in charge of young people apprenticeships and things. So I learned how to chat away to people. So we had an intranet and I was doing business and finance and I was just completely obsessed with how you could use IT to make... business faster and it was cool so cool very good giles what about you i went back to the company that i spent a year in uni you know the placement year you have so i i was a bid analyst which meant lots of financial modeling

lots of like scenario modeling dcf stuff um it was really like being thrown into the fire it was single swim what's dcf discounted cash flow modeling so So essentially on bids, you know, you're trying to work out what price to submit to government, typically for the company I worked for. Lots of cost based analysis really early.

experience with model auditing and like dealing with the big four accounting firms with with that kind of process so so yeah luckily i i swam and it set me on this path of loving excel and and problem solving, you know, mainly using Excel. And Mark. So yeah, out of, so from, I went to school to university and then after that I joined A.

Small to medium-sized firm of accountants where we did lots of statutory accounts, lots of management accounts. And that was where I think at the time we used the Excel 97 was the version that we were on. which was a pretty good version. And when I left there, whether I should admit this on the first episode or not, I...

There was a workbook that we used to use that used to prepare management accounts for a specific client. And I put a macro on there that anytime anyone opened it up, it just said Mark is the XL King. And I'm not sure anyone, I didn't pass my protector. I mean, I wasn't that evil, but equally, I don't think anyone knew how to remove that. Years later, I catch up with people and this message would pop up every time we open the workbook. Mark is the XL King. Genius.

So slightly kindred spirits there because my first job, it's a placement year in university as well, Giles. So I worked for Hewlett-Packard's finance department in the UK. And yeah, last day of the job. swapped all my bosses icons we were using lotus one two three changed what all the buttons did all the icons on the ribbon just left them to it so never spoken to him since

All right, so let's talk about what's the news, what's happened. Okay, let's get the running out of the way. Giles, what have you been doing recently? I have just come back from South Africa. I don't normally dress like this. This is my finisher's top. I've got my medal on. I promise I have taken it off at some point. It's been two and a bit weeks. So yeah, I ran Comrades Marathon, Ultramarathon, which is... uh 86 kilometers you've got to do it in under 12 hours it was very warm

It's very hilly. So I scraped through as part of a team of other people that are actually proper athletes. Unlike me, I just I just kind of wing it. And yeah, I think I came back heavier than I than I went out.

to south africa because i ate so much food uh consistently for the entire week afterwards but but yeah at some point i'm going to take this off because i'm in my conservatory and it's hot and i've got layers on because i wanted to show all this off so when we're done with this part of the conversation going to take all of this off the problem is just this is a podcast so nobody can see you anyway

Oh, can they not? It's just for the benefit of me winning Sue, really. And we appreciate it. And we've seen it. We're not putting this on YouTube as well. Did I do this just for three? It'll go on YouTube. It'll go on YouTube as well. I'll just do it for you. So that's what I've been up to. As I said in my intro, I like doing silly things like this because it keeps me mentally active. Now I've hit my 40s.

uh yeah it's quite it's quite a difference to you know spreadsheet work so given you and sue both go running are you tracking performance in excel reports power bi reports there's got to be some nerdiness involved is it just a pure did you set that question up for sue because that's like the perfect question oh well

Tell them what you've got, Sue. I've seen it. Yeah, so I have, I'm a bit like Giles. I've always run. I love running. But recently, my first marathon was with Giles in Munich, where I was going to... do a half and they were all doing marathons so I thought I'm just going to go for it and I loved it so since then I've done quite a few and

And we went to Vienna, didn't we? And met up with another friend of ours, Laura, which was just it's just such fun because you get to run around the cities. And I've just done an ultra and I've just done a marathon as well within a week of each other. So I slept. a lot but I have a Power BI report.

which i might show you next time because i haven't got it already but i use power bi to analyze my runs and i have a lovely little excel spreadsheet where i check out what i should be doing each week and Because I'm slightly older than Giles. Slightly. I have to... My aim is not to break. I do lots of swimming. Basically, I track it all using Excel and Power BI to visualize it and check miles per week and heart rate and speed.

everything. Because as we all know, if you can't measure it, you can't manage it. so yes i do and i use it for my demos so because i think everyone gets a bit fed up in the data world of contoso and adventure works and things so um well you showed it in me didn't you we did a meetup yeah munich which was awesome um so yeah you're older than me faster than me probably just tougher than me in general you take it more seriously

I do take it more so you're you're much more chilled about it whereas I'm I have specific goals I want to hit can I tell you I wasn't chilled about it 11 hours into comrade where all I could think was you're an idiot why didn't you have a spreadsheet that tracked all your progress are you though are you because we've both signed up to run budapest and

two weeks ago he went you're on your own you're on your own mate i'm not doing that again i don't know now he's ready so if you had a spreadsheet to track it and i did you a little power bi report so that you could visualize it do you think you would I don't know. So when, Mark, just to keep you up to speed, I basically retired after Comrades. I hurt so much. I said, never, I'm not doing this again.

i'm committed to guinness for the next 12 months and nothing but guinness i i'll ask me again in a month so you don't want me to go to texas with you now oh god And just to bring Mark into the conversation, Mark, how was your 5K recently? And don't laugh because I haven't even done a 5K.

hat off to mark for his 5k me and nothing nothing just just to give some background this was because the last last thursday there was a um the digital finance function awards uh for which i've been nominated for training provider of the year but it wasn't so the requirement was you had to i was it dressed to impress the problem is that i i'm too tight to buy a new suit so therefore i had

I had to lose about a stone. And I thought I would just I'm just going to start running. That's that's the way I'm going to go for it. So. Yeah, a few weeks ago, I did my first 5K and I've done a few more since then. And I mean, the awards happened. So if I wanted to, I could be like Giles now and just say that I'm on the Guinness regime for the next however long, because that seems like quite a good scheme, which requires less.

Less work. But to be honest, I've tried running several times and it's always just there was the one time I was too busy playing with my iPod and wasn't looking at the curb and I managed to fall over. face first into the ground uh which wasn't ideal and when i got home so i walked the rest of the way home with blood gushing down my knee

And I walked in the door and the first thing my wife did was phone her sister and say, you'll never guess what Mark's done. She didn't say, you okay? Literally, that was the first thing she did. So I had a kind of Apache history with running, but it's, yeah, who knows? I haven't got a spreadsheet. cheat i'm just at the minute it's just if i can get through it that's a win yeah go slow start slow would be my advice just slow it and enjoy it marco rosso does it

We've got to segue into some data. Yeah, and he was running. I went, oh, my God, you're even higher in my estimation. They said, I hate it. He said, I've worked out it's the most efficient way to burn calories. so that's why he does it so he can eat cake it's a great reason it's a great reason don't knock it yeah have you got any exercise any running any activity whatsoever

No, I'd go for a swim occasionally or a bike ride, short ones, but nah, not really. So I've got nothing. I've got nothing. I was swimming in the rain the other day. That was quite cool. I thought, I'm actually, I'm tough in the sand. The pool was heated and stuff, but I was swimming in the rain. And what was the temperature of Australia? Oh, mid-20s, something like that. Quite shitty, you know.

Okay, let's get into some data then. And let's get into some talk about stuff. Any new developments or things happened in the last, well, this is episode one, so it can be. the last year but the last few weeks that might be of interest to folks um anybody seen any good blog posts topics to talk about anybody want to shout out or should we just have a really awkward silence for like

10 seconds. I was keeping quiet because I feel like I just talked about running for ages. There's stuff from the Financial Modelling World Cup, but you tell me. Oh, go for that. Yeah, okay. Yeah, Financial Modelling World Cup. What is it?

So, I mean, the Financial Modeling World Cup. Well, so that's the name of the company. There are two competitions. There is the genuine Financial Modeling World Cup where you're doing financial modeling challenges. That's quite niche. But you've got the Microsoft Excel. World Championship. This is getting quite a lot of prominence around the world. Finals are in Vegas each year. I was invited onto Radio 2. Did you see that?

Yes, I heard you. I was chatting to Jeremy Vine suddenly decided he had an interest in this. So I was having to guide him through what he called a lookup was, which was really interesting. Anyway, so.

there are a series of battles and they're normally quite hard and um you know the idea is you can get points on the board you solve increasingly difficult problems the interesting thing i thought was this in the last battle which was a week or two ago there was actually quite a again a difficult challenge which is about summing

the sum of the digits challenge, but with huge numbers. And there's a limit to how many characters or numbers, integers that Excel can deal with. So using Python in Excel. You solve the whole workbook, which has got half an hour limit, in about five minutes. Dim Early has done a great video on his YouTube about this. if you didn't know python you were in a lot of trouble uh but it's the best example i've seen so far where

using Python in Excel offered a very, very quick route to solve a problem that you couldn't really do without it very easily. So they're allowed to use like the beta version of Excel in Well, they did. I'm not sure the case author, Michael Jarman, had kind of consciously worked with the thought that they might. But yeah, some people did. Some people were on the beta version. And if they had access to it, they solved it.

pretty quickly which is amazing because I mean I think I'm right you can even if you've got pre-made lambdas you can just copy and paste them into your workbook and then you're you're good to go so it's literally like there's There's no rules. It's like UFC or whatever it is. It's exactly like UFC. When I watch those XL battles.

yeah it's it's to be fair like it's it's it's pretty vicious at the top end of it you can use pre-built lambdas i think andrew grigelonovich is um view on this is anything you could do in the real world in excel you should be able to do in the competition the the difficulty is you've got the top top guys who are already brilliant at everything in excel and problem solving so they've now

you know taken on lambda quicker than anyone else so the gap between the top top guys and everybody else has grown um i still think it's a fantastic way to test yourself if anybody hasn't heard of it go and check out The Microsoft Excel World Championship. It's $20 to enter a round. It'll be really hard. You'll do really badly. And you'll definitely improve if you keep trying.

There you go. Do they get a prize? Do they get a prize? Yeah, there is prize money. And I think year on year, it'll grow. So I think it's about $25,000 at the moment, the prize pool. Like I say, I think it's got real legs, this competition. And it's an esport. It's a genuine esport. So you think of what...

how many millions people can make playing Call of Duty or Fortnite or whatever. Why not make millions from solving problems quickly in Excel? It's our dream, isn't it? Yeah, yeah. Because that's the one that's like... televised on ESPN stuff isn't it yeah yeah so I tried I watched an episode a little while back and it was just going so fast and the sort of screens were zooming in and out I was just like I'm not really sure what's going on

I think it's one of those ones you've got to get into a little bit. You do. And bear in mind, people like Dim and Michael Jarman and Andrew Nye, who's one of the top guys as well. These guys have been competing for... years and years and years so not only are they very intelligent and very good at excel they're also very experienced battle pros so so

I always say to people, you know, watch these guys and take inspiration from them, but also don't compare yourself to them because if you do, you wake up every day for the rest of your life being immensely disappointed in yourself, which doesn't do any good. Is there any chance you could use some form of brute force attack on the answer sheet and just kind of spam it with every single number? Your head's still in UFC mode, isn't it?

It's like I can't boot them using standard rules and I need to find another way. Python. This might be a nice segue.

out in Regex. Regex will be another tool that I think will be useful for the battle people coming up. Oh, okay. Yeah, the Regex thing, you know, it's not something I've ever had any experience of. um so and i've never really i don't think i've ever missed it so it's one of those tools that you know it's used for it's in insider version currently of excel good for extracting data yeah from you know splitting strings of text apart and grabbing something finding patterns and all that sort of stuff

but not something that I've ever had much of a need of. So it'd be interesting to see how it develops. I've used it in R for analyzing questionnaires. so you know when people do free freestyle questionnaires putting out certain words and such like but um not within excel I guess the only thing you would do is if you're comfortable, very comfortable in Excel and you would download a CSV, it would be using the regex on one of the columns, I think, without then having to go to another language.

I think that's where the use case is seen. So quite a bit of excitement about it. Any thoughts, Mark? I used Regex about 20 years ago when... At the time, I was thinking about starting a website development agency. I didn't, but we built our own content management platform, which had people could log in and they definitely needed to verify email addresses and other things.

so there were there were patterns that we would just use uh that would check whether various email addresses were were valid but apart from that i've never really understood how that works all those characters you just look at it and it's a complete mess i think most people at the minute are just suggesting that you just go to chat gpt and ask it to generate some some regex because actually the to create some of those test strings is um

like it's a whole language in itself and it started like the 60s or something so it's kind of a it's been around for a long time but it's it's just quite hard to to master and to get it right so my plan would be chat GPT or I'm sure somebody on stack overflow has asked my exact question at some point so um that if I ever need it that will be my plan yeah

uh chandu um chandu shared a really so there's a really interesting day or two when lots of the mvps together or in what looks like quite a coordinated way released new reject videos chandu is the first person that i saw post that said hey look this is great but also like if you've got bad data regex isn't the fix it might help you extract things in a cleaner way but you still need to go back to source and ideally sort the data out which i thought was a great point to make amongst the hype

yeah it's the context of which you're pulling the data out from so if you're looking at sentences there was um There was something on, I don't know, one of the memes saying commas matter, you know, no comma more wine or no more wine.

So it's the same with rejects, you know, when you pull something out and you're trying to analyse the meaning or the sentiment behind it or what it's trying to convey, which is what I had to do with the sentiment analysis, is... is context you know what what are the what are the phrases that it's used with so it's sometimes it's not just about one word it's about the whole thing so

But then it can be useful, like you say, for validating emails or looking at bank numbers or sort codes or something. That's, I think, where it's, you know, a predefined field. Does it meet that criteria or not? No, well, only just to play about and see what it was about and sort of, yeah, try a few things out. It feels like it should be a Power Query feature rather than an Excel function.

I mean, it's because it's about extracting, you know, you're trying to clean data or extract something from your data to get it into a format that you can use. To me, it feels like it would sit better in.

power query because that's what it's for but um whether whether we'll get it we'll find that i think there's a few people who have tried having regex in excel i think some people have ended up using some kind of javascript derivative or something I'm sure someone's asked me that and I've searched for it at some point there'll be some power query hero who's already sold that and it's not me but some some some i'm sure somebody has um

I think Brian Julius, who's in the Power BI world, he's used Power Query, but again, using either R or Python as the module within Power Query to use regex in that way. But it does seem that Excel is... bringing giving much more programming functionality towards data cleansing with the lambdas and regex yeah with with excel now i think

They changed it all under the hood so that an Excel cell can contain all sorts of stuff now. So not just a number, it can contain an image, it can contain all this other stuff. and the grid is sort of not so limited anymore. And they're just sort of, I think like the engineers under the Excel team are just lighting up with like, oh God, we can do that now, we can do this now. And I think.

Some of these things will fall in the gap between, you know, what's cool and what's business practical, what's really useful. And other things will just, you know, turn out to be properly useful, but they're just trying a whole bunch of things right now, which is pretty exciting. It's all happening so quick. I think he's brilliant.

Sorry, because I think Excel, people are programming if they're in Excel. If they're using an if statement, they're programming. And so they don't realise that. So when you say to people and you're training them or talking to them, actually, you are programming. Am I? And you're like, yes, you are. You're thinking that way, which is fabulous. But from a data perspective, having...

all of the functionality that you can do in Excel. It's like, please don't use Excel to load into Power BI because you have to be quite careful. Make it a table.

please i'd be i'd be interested to get your all because we're all involved in sort of training in one way or another but what i still see you know with full stack is Most people are nowhere near the point where I would suggest they start thinking about worrying about Python.

or regex or anything like that and and actually we do get a lot of people that come in going i really want to learn python then you dig into you know why you know you're still using vlookup or you don't know what an excel table is so i don't know whether you have that shared experience where the new flashy stuff is great like as an excel geek i love it but for 90 plus percent of people

in this Excel space, I would probably try to hold them back from all the new stuff for a little bit. For me, I think the most important thing that Excel user needs to know it's about how you structure data properly because then once you get that right everything else can flow out reasonably easily so until someone's got that completely sussed I would say don't even

So Python and Excel is really about the visualization piece because they haven't added a lot of the other libraries around automation. So it's really about that visualization piece. So if you haven't got... data in a good structure and your visualization is going to mean nothing so from what i think it's for excel users for excel users it's primarily about how do you structure data so then you can work with it

in the best way so it's and that's why power creates an important tool because that's the tool that gets us from here's my input to here's some data in first normal form or whatever structure we're going with but that's that's the most important thing so i don't think the so i'm with you jails i don't think that python or projects or those things are ever going to apply to this because it's that's just not

That's not what they need to fight with a VLOOKUP because they haven't got their reference number in the first column. it doesn't matter what what shiny stuff you put on top it's not going to make any difference yeah no no arguments for me it's sort of still people on training courses doing power bi's sneak a bit of excel in there show them tables in the next look up

And it's just like, oh, what are those things you're showing us? It's just, you know, it is the core elements. I think if you know tables, XLOOKUP and pivot tables. in in the curve of things you're you're advanced you're uh you're you're more further ahead than you would think so you know don't feel like you're falling behind if you don't know some of the advanced stuff exactly the core things yeah absolutely yeah yeah the other the other thing just in terms of bits of news

I saw a good blog post from Kurt Buehler. He always does good stuff. He's brilliant. He's brilliant. Sort of Power BI related type stuff, but just did a good video on what makes a good report about, you know, what to present.

sort of you know have this it was this concept of having in the first three seconds somebody should get a number an impact a bit of meaning you know when they look at it and then after the sort of 30 seconds you get a really good sense so when you're building a report in excel or power bi it's that sort of impact thing i think we see a lot of reports produced on linkedin and other places that are quite shiny and nice but lack lack a lot of um

know there's not any insight in them for me for me i say this jokingly but i'm sort of semi-serious the ultimate dashboard is a is a green button that says everything's okay yes yes no i would agree because people people are only going to dig in deeper yeah to the reds you don't make somebody work really hard to work out that everything's just all right you know you should be producing that level of you know red button green button and then

That leads on to my thing. Can I tell you what I've just done? I'm very, very pleased with it. I love Tabular Editor. So those of you that don't know, Tabular Editor is an external tool that attaches to Power BI. And what it does is it allows you to do things very quickly. so um and because i work for myself uh time is important so i've just get me not a programmer so i've just written a little c-sharp script that creates a measure for me and a measure does all the calculations

stuff in Power BI, which will format, I can write a measure which will format everything to be either green, amber or red. So I can apply it to the conditional formatting in Power BI. Let us go. do that and it gives me a little measure and then i drag it into power bi

And it goes, if it's green, if it's amber and it's red. Because like you, keep it simple. My users have got enough to do without having to try and decipher what the report is doing. And quite often, it's like, if it's green, don't even show me. Just show me the red so I know.

which ones to focus on so um i'm like that big win today a little bit of so i i guess we may have all experienced this as well this thing of wasting time automating things manually in five seconds yeah i know i know but it's it's it's time invested isn't it Absolutely. It took me probably about an hour or so of trying different ways and things that I could do to automate it. And then I was like, right.

let's try again and so it's about i don't know five or six iterations and i've got it now did you start that story saying you're not a programmer sue because i mean just perception wise like to me you're definitely a programmer But you don't want to give yourself as one. No, no. I would say I'm a self-taught programmer and I don't consider myself one because I don't know how to use Visual Studio Code very well. Or GitHub.

But I'm getting better. I've started to use Git and Visual Studio Code now because of all the stuff that's coming in Power BI, which is very exciting. Yeah, you're a programmer. I was going to say to me, you're definitely a programmer. I'm a grid XL person that's dipping his toe outside. You're a programmer.

So this is the first episode, so it's probably going to be quite a small number of people listening to this to start with, although knowing Wyn's little YouTube page, it may not be quite so small.

i've got sons and my first son is called ross and i have to say i'm slightly ashamed but also slightly pleased that when i was thinking of what name to call him i love ross because it's poldark which is just like cornish and you know definitely de maurier and everything but there was an element in my brain that said actually you're going to be saying this name repeatedly for many many years

constantly so i went for a nice short name to save time it wasn't the pure reason but it was a reason because i thought oh that's going to be you know well i'm going bartholomew thinking ahead thinking all that planning that's where my brain works that's that that is automation you see you thought if i go for a long name you have to do that thing over and over again but if you go for a short name

like a keyboard shortcut that is set it up right in the first place so you can see why i'm slightly embarrassed but also slightly proud it's good it's good um right are there any other news or should we just get into the main topic for today was that not the main topic what was that 35 minutes and we're not on the main topic listen on the intro this is should i talk about running more let's go for the main topic come on

See how many times you can fit running in in the first few episodes. I'll do a little connect. Stream the chat and then use regex to pull out and then sentiment analysis for each person. how negative or positive they were that would be quite cool excellent i really like that idea she's not a programmer but she's gonna do that main topics sort of ai chat gpt co-pilots what's all the fuss about yeah is it is it overhyped is it good

Where are people at with this? Who wants to start? Anybody got a preference? Mark, do you want to? Oh, Giles has got his hand up on the screen. Well, you asked the question, so I just thought I'd go visual. Go for it. Yeah, go visual. a a kind of a follow-on from what i said about uh basic skills anyway uh yes i hear lots of hype about ai and copilot most of what i've seen for copilot in excel hasn't blown my mind but i'm also very much in the mindset that this is only going to get better

I mean, I don't think you can say AI is a fad or anything like that. Like, what was the fad that still hasn't worked? I've got a PS. I've got a PlayStation, and they tried to get the, what's it called? or like the virtual reality stuff like the big headsets so you have to kind of set things up across the corner of your room and all of this that's a fad ai is not going anywhere it's only going to get better so

I kind of still tell people that I speak to, like, yeah, you've still got the fundamentals. AI is not just going to do everything for you yet. But I think it will be an absolute game changer at some point. That's me. Yeah, I would say in terms of creative, I'd say creative activities, right? So anything that, for Copilot, anything that's in Word or Outlook.

or powerpoint it's there seems to be a reasonable use case for it and i use that from time to time and the chat gpt um it generally tends to be around more creative type activities which when you come to excel And you don't want a creative type activity. You want something that is if you're if you're asking it to write a formula or something, you want it to be accurate and optimized and and to give you the same result every single time. So that's where I think.

that Excel has got a challenge where in the same word if you say I'll provide some text on this if it And if the next time you wrote it, it gave you a different sentence structure, right? You wouldn't care. You'd roll with it. You'd read it. You'd change it. You'd amend it. No problem. But in Excel, because it's either it's right or it's wrong.

And if you get it wrong in your first formula, everything else that flows after that could also be a problem. So I think the I think the use case is the challenge inside Excel. Now, inside ChatGPT, if you're asking. If you're asking that to generate a formula for you and then it provides you with your description and you can or you can ask it to say, describe this to me. It's I almost feel like that's a that's a better.

environment than excel because in copilot for anyone who hasn't seen it or hasn't used it you you type in our add a formula that does this and it inserts it automatically into your into your table the problem is is that once you've got down various layers you then can't go back and say actually i didn't seven formulas ago i didn't mean that i meant something else it's kind of it's iterative at the moment so i kind of feel like there's just a challenge in terms of how people

implement it and when it comes to when it comes to things that people don't know or understand so if i ask if i ask chat gpt to generate some text that i might use so prepare an email for whatever I'm going to read that email and I understand what that email is and I understand the nuances of using various words and how various words come across as quite strong and I might decide to soften it and change it.

And I'll think about my audience and who they are. And I don't want to use that term because they'll think this. And so all of those nuances we understand in terms of language. But when it comes to Excel, it's going to pump out a formula and it might give me an offset.

which if you don't know, that's then a volatile function, which if it's depending on where it is in your calculation chain can cause your formulas to become quite slow. But you don't know that. So there's lots of nuances there that I think we end up missing. And I went in terms of words, I can already speak or hopefully, hopefully I can speak English and I can read English. Therefore.

I've got some knowledge, but in Excel, people are dealing with things that they don't understand. So I think there's potentially just a house of cards situation. So there's quite a lot of risks there. At the moment, I'm a bit torn as to whether it's... a good thing or a bad thing? I know that's the way things are going to go. I know that's the way the future will be. But is it helpful? Is it not? I'm just not sure at the moment. I love it.

I was doing some training yesterday. I was showing some people some code. And I said, do you want me to give you that and explain it? And they went, yes, straight into chat GPT, comment this out for me, explain what it's doing, put the code.

in and then it's out is really nicely explained and I wanted to do some dummy data for something so I took a picture yeah took a picture of snipping tool of my data model and said prepare me some dummy data for this i need it to have this this and this and that and then it did it and And quite often I find sometimes starting something like text wise, I sort of know what I want or I'll do it and then I'll say, right, can you improve this? Or so I always find.

When I'm using it, it's a two-way process. Like you said, exactly like you said, Mark, you have to know what you want it to do and to be able to validate. whether it's right or wrong, because quite often it's wrong, and I'll go back and say, no, you've got that wrong, thinking, oh, yeah, my bad, Sue. It does, it goes, my bad, Sue. So, you know...

You have it's a two way step. But what I would say is it won't fix things for you, but it is a really helpful member of your team if you use it in that way. Yeah, I think it comes from a place of having expertise to start with.

you can understand it but yeah but it's that lack of expertise that concerns me but i think when you've almost got it as your as your google replacement yeah i'm just i'm just using it on my phone constantly and searching for stuff using it and things like that just in a presentation actually on the romania power bi user oh did you do it for christian yeah yeah so yeah so that video will be available somewhere so using it for excel

power bi and stuff and then i did i tried copilot briefly at the end and sort of and copilot's a good name right like i said it's good it's good yeah um but The ChatGPT stuff, I put a really complicated formula in there. And I'd already built a prompt that sort of said how I like my formula to be reformatted. And I told the prompt to, can you also just check if there's a better way to write it using let.

So this long gnarly formula is then simplified using let and it still works and I can validate it and it still gives me the answers. But I understand the let.

went into chat gpt said can you explain this formula to me sorry went into copilot in excel the thing that's i think it's 30 bucks a month so went into copilot clicked on the formula said can you explain this and they went i can't find a formula it's just like oh come on it's there and that's the experience you know chat gpt blew me away the very first time i used it and then i did a follow-up question and it got it did more again and it was just good

All my experience as a co-pilot have just been, oh, it doesn't do that. Oh, it doesn't do that. You know, I opened up a Word document and said, can you just suggest a way of me referring to this or rewording this paragraph? it says oh i can't access anything in this word document okay it's copilot for word it's just those sorts of things just really like the the expectation and the delivery

There's a massive gap. And expectation-wise, you know, I still think it might end up here with Excel, but what I hoped it would be was if you've got a really... large data set if you pointed copilot at it especially when you think of all the new stuff in excel you know python regex functions whatever else if copilot could do the job of kind of

assessing all the tools that it now has available in modern Excel and doing the best job of finding trends and insights from that data using all the tools that are really hard for you to do manually. would have blown me away and i think i read somewhere on some forums that python it was supposed to be able to lean in on python but apparently it's it's it's not there i have to admit i haven't tried it but

it like you said when it's like the difference between the expectation of what it would do and what you actually see it being used for is is quite significant yeah i had a call as well or an email from somebody the other day saying their boss had asked them to look into implementing AI in Power BI and Excel and things. And it was just like, why?

you know it was just like we need this this sort of knee-jerk reaction of we need ai for stuff for stuff and like as we mentioned earlier you know people are still down in the crappy data Yeah. If you can't get your data in, you know, the investment will go into AI versus the investment will go into actually teaching people how to do the basics in Power BI or Excel.

I think there's just a massive disconnect there. That's the stuff you should be sorting out. And I know I'm biased because I deliver training. So I'm obviously going to say training is the right idea, but it is. get your staff up to speed with the latest features and what they can do with the core elements and how to do it properly. And then the other stuff will be, you know.

If you're chucking garbage into AI, you're going to get garbage out. So it comes back to this, you know, a lot of it comes back to tabular modelling and automation. Somebody had a poll on LinkedIn a few weeks ago about whether... And I don't know how many people responded, but about whether time would be better spent learning AI or learning basic Excel skills.

And I wasn't sure which way it was going to go, but it ended up being, so there was kind of strongly agree, agree, and, you know, the five options. But between agree and strongly agree, for example, it was like 92% of... People thought that basic Excel skills would give you more benefit than being really good at AI. Learn Power Query. Learn Power Query would be my answer.

to automate everything automate everything clean everything and do it once the the way i the way i see the you know the the chat gpt stuff its restriction is that that's security you know the security around the sensitivity around your data so there's there's an element there but if you just put code in like formatting code excel formulas there's not really a sensitivity issue there but if you're uploading data then there obviously is

but that's where i would hope that copilot in that secure environment you know you can just freely do whatever you want i just wanted copilot to work like chat gpt probably will at some point it will i think you know usually the same model in the back end as far as i'm aware but it's just i guess as well yeah there must be a layer around it to do with the tenant settings and security that maybe chat gpt doesn't have and the complexity of actually accessing the excel grid

you know, is just insane. But I would be totally happy if there was just the chat GPT functionality in the side panel in Copilot. Just that, let me copy paste a formula in.

and explain it don't even interact with the grid as mark was mentioning earlier you know if if that's what would give me the best results so i don't have to go off to copilot to do stuff that would be awesome The best use case I've found for Copilot so far is you put a table on your grid and then you type out into the search box, can you provide some VBA code to do whatever?

and it doesn't even press the grid you just need a table there so it turns on the feature and then you can start talking with it about other stuff brilliant because auto save has to be on as well in in copilot in excel which i still can't get to grips with autosave to some extent it's all right but i know that that whole thing of you know every change you make just happening instantly so i can't sort of close out with something without having saved it

it just still still rubs me the wrong way but maybe i'm just a bit old in the tooth long in the tooth doesn't doesn't uh polarize opinion as much as uh automatic or um manual calcs if you get into a financial model of well that that's a topic that will get some heated answers for you should you have your excel in automatic or manual calc mode

these are the folks that you know legend says break their f2 keys off their keyboard or the f1 key the f1 key in case the accident i don't know if that's actually true but it's a good mean no no nobody wants manual calculation mode it's just because sometimes your model gets so big that you need it so surely nobody wants it it's just the fact that you're stuck with it

so i would say always go automatic until it becomes so big that you've got no other option and then you should be using power query yeah i'm with you but lots of project finance modelers would strongly disagree with you Hopefully they're not listening. Hopefully they're not knocking down my door. I mean, at the start, I told them where I live. Ah, rookie mistake. I've got no chance. I, well, didn't refer to that.

So how much of it, and from the Power BI world as well, I don't like this term general availability for co-pilot when it's only available if you pay five grand a year for... premium. So that terminology, I know it's technically correct. It's generally available, which means that if you have a bug with something, the support will answer and try and help you with the issues.

rather than being in preview when they won't but it just goes hey it's all generally available which means oh but only if you're on the premium if you're on premium and that sort of you know is actually it genuinely costs more when people run copilot. So, you know, they're paying for the processing power. They're paying for the chips and the compute power. So I can see that it is a genuine feature that they need to charge more for. But I just I'd like a bit more open.

well chat gpt's stock you don't have to pay now do you for um part of it yeah you've got access to more of it it's just um if you want to upload images and analyze them you have to have the paid version if you want to create your own

little prompts and save them. You have to pay for it, but you can now access the prompts for free. So yeah it's it's pretty cool it's just it's growing and growing and it's just going to get faster and the models are quicker and your code gets written quicker but like you say if you don't know what you're doing you're just writing code faster than doing the wrong thing yeah i run some i put

two reports together or two data sort sets. One was like a dimension table with customer and product, another one with sales. Build me a chart and it built me a chart and it was right. Okay. Then I said, show me who's the top selling customer and it was right and I said give me that chart for the top selling customer and it looked right and I have no idea why because I have no idea what it did under the hood to generate the numbers

yeah so there you know it's still it's it's you have it's not good enough being one out is not good enough you know wrong wrong is wrong when it comes to the numbers yeah because you don't know what it's gonna be if it's someone else so yeah yeah so trying to assess that is just really tricky but getting a lot getting a lot better

So any other thoughts? Yeah, one thought on this. If you can start to rely on it getting the right answer, I just wonder whether Copilot will be... of most value to people that don't like excel do you know what i mean so if you're just a user who would rather do anything in their day other than be in excel they have to use it because they're told they have to use it

i wonder whether copilot is going to be most valuable to those people because actually it's probably going to be best at lots of the fundamental tasks that we i think the four of us would think of as very easy um so i i don't know it's just a thought that for the for the probably vast majority of people that don't wake up and go to bed thinking about how they can get better at excel like us maybe it's best for them because it's going to do a lot of those usury tasks

quickly and hopefully reliable hopefully reliably for them yes but i think it's still it misses it misses it misses the piece It misses the piece about the data, even though it's done a good job in the past with the validation and also the restructuring of the data into the right layout and things to give you the answers. And if... you know how how accurate does it need to be like people make mistakes okay so

We still trust people to give us the right answer most of the time, but can we guarantee somebody's going to give us the answer most of the time? So where is the tipping point where we trust Copilot? How many times does it have to get it right? That's why I caveated that with assuming that's right, which is a big assumption. But so in my head, right, if you just had a massive data set.

it's in a you've you've copied and pasted it into excel and you just point to copilot you say analyze that and stick it into a five page powerpoint for me if it just did it nine five percent decent job on that for the average person maybe that's the best use case for copilot for the foreseeable future assuming it does the right stuff um i don't know it's just a thought that popped into my head

I think that's where it falls down is that you know you can just imagine the you only have to hear about somebody else having a bad experience still one two people three people four people and you think you know the news gets around And it's just wrong. Imagine, you know, an Excel formula just giving you the wrong answer one day and everybody goes, oh, yeah, Excel gave me the wrong answer. Excel gave me the wrong answer.

Oh, that's not good. Power BI, you know, the report was like, there was an article, I don't know if you saw this recently, Sue, with the article saying that Power BI was a security risk. It was just a complete fluff piece by somebody. And essentially, there's a publish to web feature with Power BI, and you can share your data, but you haven't given someone access to the underlying data set. There's no risk.

they saw it as a risk and said oh there's a big security and it just went you know just went crazy there was something similar with vba a little while back because it was a security risk which to some extent it can be Just the hysteria. Yeah. I think people are just, you know, like you say, are saying that and then they're sending out emails to the front and center, which are in it. So it's just frightening everyone.

yeah i need to um are we ramping ramping wrapping things up for today but we've done a good 56 minutes for episode one I know, it's good. I've got a call in. How many main topics were there supposed to be? Did we just barely finish one? Another one, that's all right. We can roll it over. We just talked about it.

yeah perhaps if you do a podcast and ask people to say what they want us to to try and keep us on track so yeah yeah we'll be we'll be putting this on um various podcast platforms we'll be on youtube um you you'll know about one of those because you'll be listening to this otherwise you won't have a clue what i'm talking about and we'll do another one in a couple of weeks time so our plan initially is to do a fortnightly episode

And we'll just see how it goes. I enjoyed chatting to you guys. This is our first episode. So thank you. We got through it all. We got through it all. No technical issues. I thoroughly enjoyed myself. Thank you. For a long time. And Guinness. It was a pleasure. I might have to go for a run. Hey, I had a pint of Guinness in Dublin. It is where the best Guinness is. It's the best.

I did enjoy it. Maybe we should do a challenge. By the end of this season, we will do a live episode where we are running and drinking Guinness at the same time and having a conversation between the four of us. Two of you could run. Two of us could do Guinness. Yeah, you two do the Guinness. I'll do the running in the conversation. And it's been a pleasure, Wyn. Thank you very much. It's been a pleasure. Excellent, everybody. Lovely to chat to all of you.

Okay, thanks everybody. We'll catch everybody in the next episode. See you next time. got these tables messy as can be columns wide and jungle nothing's clear to see precise gonna pivot these fields oh baby that's nice you gotta unpivot make my data sing from those cluttered columns let the insides ring transform and tiny now it's all in use With the on pivot, baby, I shake off those data blues Tell