[Kevin 00:07]
So Zoe and I recorded a whole load of mini-interviews at the MVP Summit, and we thought rather than build those into a few larger episodes, we'd release those to you in short little snippets like this. So enjoy a set of our interviews live from Seattle with various people from Microsoft and MVPs. I hope you enjoy them.
[Zoe 00:30]
Hello, I'm here in Seattle for MVP Summit with Rishi. Would you like to introduce yourself?
[Rishi 00:34]
Hi there, yes. So Rishi is up first. So I work in the Global Intelligent Data Platform practice at Avanade, which is structured the same way as Microsoft. So Avanade structure is also the same way. We cover everything data platform. So Databricks, Fabric, and all the other platforms that form part of IDP.
[Zoe 00:51]
You are an MVP. Are you a Data Platform MVP?
[Rishi 00:54]
Yes, Data Platform MVP.
[Zoe 00:55]
Brilliant. Okay, so thank you for joining me today. I've just got a few quick questions for you. First of all, what does CoPilot mean to you?
[Rishi 01:03]
So CoPilot, you know, when I first started using CoPilot with ChatGPT...
[Zoe 01:07]
Which CoPilot? I mean, not.
[Rishi 01:08]
Just ChatGPT, even Excel CoPilot, and all.
[Zoe 01:11]
The.
[Rishi 01:11]
Things taken off as CoPilot, I started to think of it as automation, because that's my background that I'm used to. And I started to say, PowerPoint, can you make all the fonts in this PowerPoint deck consistent? No, sorry, I can't do that for you. Can you apply branding to this?
[Zoe 01:24]
It's like the consultant's dream. Yeah, yeah.
[Rishi 01:26]
Yeah. No, sorry, I can't do.
[Zoe 01:28]
That yet?
[Rishi 01:28]
What can you do? But then I think you start to realize that it's not about automation. It's actually about having that creativity. And from a data perspective, that's actually more useful than automation, because you could do Excel, VBA, and things like that. And yes, you can get things to write code for you. But actually, what's more useful is thinking about, well, how do I need to design the structure of my tables my data to be able to make get the insights that I want and actually humans are quite bad at doing things.
[Zoe 01:57]
Like that.
[Rishi 01:59]
We can't connect all those different dots and we can't go out to the internet and think about outside of my financial process I've got this kind of data what type of analysis could I do and if you actually if you ask Power BI copilot that question it gives you.
[Zoe 02:13]
Some.
[Rishi 02:14]
Really really good answers. So that's what.
[Zoe 02:16]
I'm thinking about recently. Brilliant and that leads me on really nicely into the next question which is Can you give us examples of how you've used CoPilot in a way that has delivered value to you, or that has been a benefit?
[Rishi 02:31]
Yeah. I think it's really thinking about actually some of those non-technical things as well. Say for example, in Power BI, everyone focuses on the tech and the tools, and this feature, and this feature's a tablet, this feature's a Power BI. Actually, the most important thing to get the right adoption and the right usage of your Power BI reports is to understand the requirements, to understand the business questions that a user wants to answer from a report, and then to be able to design a data model and the storyboard and the visualizations to be able to do that. And actually, as I said, that's not something we're naturally great at, especially the technologists, right? So actually, maybe that's where CodePilot can help the technologists and say, right, who's the audience for this? You can't be offering this to all people. What are you trying to build with this Power BI report? What kind of questions? What's the problem statements? What are you gonna do with that information? And then to say, well, here's the kind of story you can tell with the data that you've got. You've got these fields and these columns. You could do these calculations, you could do this type of analysis. That's what I think will be really useful.
[Zoe 03:29]
Yeah, brilliant. So I haven't used Copilot in Power BI, but you've made it sound really powerful.
[Rishi 03:35]
Yeah. It's specific. Power BI Copilot is really good at coming up with the suggestions, but I think it's thinking about it in a broader sense. Yeah. It's thinking about, can I use ChatGPT and feed in a persona, feed in a set of problem statements, and let it come up with insights, and then tell me how to structure my data to feed those insights and what logic I need.
[Zoe 03:51]
Yeah, so it feels like we're at the start of a brave new world, and with that in mind, like, is there anything that you're worried about in this new era of co-pilots?
[Rishi 04:01]
I think probably too much focus on tools and tech.
[Zoe 04:04]
If I'm honest. And it's technology, so that's probably hard to say, isn't it? Yeah, no.
[Rishi 04:09]
Absolutely, because, again, people are thinking about what can copilot do for me, what's the feature that this has versus Google Gemini or something, or a Power BI has versus as Tableau Pulse or something like that. And actually, it's less about that. It's much more rethinking processes, rethinking how you approach a problem from start to finish. Like, imagine you were just starting a brand new company and you had to get to this side of output, that level of insights. How?
[Zoe 04:34]
Would you do it? How do I get to product market fit when this is the problem statement I'm trying to solve.
[Rishi 04:39]
This is the problem statement, whereas the way everyone's thinking about it is, I've got to do this process, this month-end process, or this process, how could CoPilot help me do it? Well, maybe it can, maybe it can't. Maybe it understands your process of data and business well enough, maybe it doesn't. Maybe you need to do a lot of work to get it to that point. But actually, rethink that, what are you trying to do? What have you got to start with? And help copilot guide you. And I think training will be a really big thing part of that.
[Zoe 05:04]
Because I'm.
[Rishi 05:05]
Really hoping copilot will start to become a teacher a bit more than it is now.
[Zoe 05:09]
Yeah, I mean, it's definitely a colleague, isn't it? So in the Office copilot, I can use it as a colleague and ask it questions and for guidance and ideas and help. And I'm assuming you can hopefully do the same a little bit in the data world.
[Rishi 05:25]
Yeah, but actually, tell me how I can do something.
[Zoe 05:29]
Yeah, yeah.
[Rishi 05:30]
I could do it, I'm actually happy to do it, but I don't even know where to start. Especially with something like Fabric, there's a million and one ways to do.
[Zoe 05:36]
The same thing.
[Rishi 05:38]
You can look at my data, what's gonna be better? I can give you my data volume so you can see it.
[Zoe 05:43]
What's the best, most efficient?
[Rishi 05:44]
What's the most efficient way? What's going to be the best use of my capacity units? What's going to be, what's the code to do?
[Zoe 05:49]
It then? Yeah, yeah.
[Rishi 05:50]
Give that to me or give it to me a starting point and guide me through that process, all the whole analytics process. That's where I hope it's going to get to. The problem at the moment, all the co-pilots are disjointed.
[Zoe 06:00]
So. Yes, yeah, yeah, yeah. We'll get there, we'll get there. But also, yeah, go on. That's a really good point for the last question. If you had a magic wand, What would you like to see next in this era of co-pilots?
[Rishi 06:13]
Yeah, so, I mean...
[Zoe 06:15]
Get rid of that disjointedness.
[Rishi 06:16]
You've got to be careful what you wish for, right? Because I asked today the question, actually, in one of the product groups, and I said, you know, look, you've got all these different co-pilots. In Fabric, you've got Data Factory copilot, Power BI copilot. They don't seem to talk to each other. But actually, if they did, could they not feed the information to each other? Could you not say, I recommend you structure your data like this, and by the way, I'm going to pass that onto the Power BI copilot so it knows how your data is structured. And I guess the challenge with that is that's very hard to have that overarching Microsoft copilot. And you start to get individuals to Skynet very, very quickly.
[Zoe 06:48]
Yeah.
[Rishi 06:49]
You've got one master copilot that rules them.
[Zoe 06:51]
All. Yeah, I think as well that the human in the loop is really important because it can make the recommendations but I personally still think that before it takes action and passes that on, it should be an approval. Like it can say to you, We think this is what your data needs to look like. Do you agree? And then you send it to the Power BI copilot.
[Rishi 07:13]
Yeah, but it's two things. Number one, it needs to guide you.
[Zoe 07:16]
Through that process. Yeah, yeah, yeah. And that process needs to be possible. Yeah.
[Rishi 07:20]
So the co-pilots don't necessarily need to all just link great together and seamlessly do everything for you because that's not gonna help you. But I think the other angle is, give you the opportunity to provide that business context. So what does this data mean? What am I trying to achieve? And by doing that, A, you're involving the human, but you're actually using the human for what humans are good at, which is what they know.
[Zoe 07:39]
Yeah, brilliant. Well, thanks so much for joining me. Thank you.
[Rishi 07:42]
Thanks for the opportunity.