In this episode of Alter Everything, we talk with Ian Barkin and Tom Davenport, authors of 'All Hands On Tech: The AI-Powered Citizen Revolution.' They discuss their motivations for writing the book, the emerging role of citizen developers, and the democratization of data science and AI. Themes include the evolution of low-code/no-code tools, the importance of governance in deploying AI, and future implications of generative AI in citizen development. Listeners are encouraged to register for the Alteryx Inspire 2025 conference and can access free chapters of the book via the show's website.
Panelists
- Tom Davenport, Distinguished Professor @ Babson College - LinkedIn
- Ian Barkin, Founding Partner @ 2BVentures - LinkedIn
- Megan Bowers, Sr. Content Manager @ Alteryx - @MeganBowers, LinkedIn
Topics
Transcript
Ep 181 The AI-Powered Citizen Revolution
Introduction to Alteryx Inspire 2025
[00:00:00] Megan Bowers: Hey, Alter Everything listeners, we wanted to let you know that you can join fellow data lovers, analysts and innovators at the Alteryx Inspire 2025 conference. It's the analytics event of the year, happening May 12th through the 15th in Las Vegas. Head over to alteryx.com/inspire to register now. We would love to see you there.
Meet the Authors: Ian Barkin and Tom Davenport
[00:00:27] Megan Bowers: Welcome to Alter Everything, a podcast about data science and analytics culture. I'm Megan Bowers, and today I am talking with Ian Barkin and Tom Davenport, the authors of the book, all Hands On Tech, the AI Powered Citizen Revolution. In this episode, we chat about themes from their book, the emerging role of citizen developers and how they use Alteryx, and how Tom and Ian are seeing companies operationalize ai.
Let's get started.
Tom, Ian, it's great to have you on our podcast today. Thanks so much for joining. Could you both give a quick introduction to yourself for our listeners?
[00:01:05] Tom Davenport: Go ahead, Ian, you're alphabetically first.
[00:01:08] Ian Barkin: Sure. I will embrace that alphabetical dominance. Yeah. My name is Ian Barkin. My career has spanned both outsourcing and automation.
For the last 12 years, I've been helping enterprises interpret and then apply and manage the types of automation that were available. Throughout those 12 years to digitize operations, front, middle, and back. I did that through entrepreneurship and a startup. I had, I've done it through courses and trainings and other content I produce as well.
[00:01:37] Tom Davenport: And I'm Tom Davenport. I'm a distinguished professor. I think it's my official title at Babson College. Faculty director of a new Center on Tech and entrepreneurship at Babson. I'm also a visiting scholar at the MIT Initiative on the Digital Economy, visiting professor at the University of Virginia Darden School at the moment, and a senior advisor to Deloitte's Chief Data and Analytics officer program.
[00:02:08] Megan Bowers: It's gonna be hard to fit all that in our episode description, . You do a lot. Yeah.
Exploring 'All Hands On Tech'
[00:02:13] Megan Bowers: I'm excited to learn from you both during our time together here, but one of the reasons why we're talking is because you guys wrote this book called All Hands On Tech, the AI Powered Citizen Revolution. And so I'd love to hear about like what trends you were seeing in the industry that you wanted to spotlight.
What was the motivation for writing this book?
[00:02:34] Tom Davenport: Let me start and then, yeah, I think Ian has a slightly different motivation, but I have mostly worked in the realm of data science, analytics, big data, ai, and I was particularly interested in what was happening to democratize those activities and saw a fair amount of that going on.
I'd written a little bit about it, and then a couple of years ago I'd written a piece on . Low code and no code, which I knew less about, but it was in Harvard Business Review, and I think the first thing on that topic, I don't know, Ian and I started talking and he was interested in it as well, but I'll let him tell you why.
[00:03:11] Ian Barkin: Yeah.
Tom and I have
known each other for probably about 10 years now. We met back when I was running that RPA consultancy. I was fascinated by this inflection point that we were rapidly finally approaching. There had long been this aspirational comment around how technology was getting easier and more people could actually use it to digitize their day-to-day tasks and work.
But much of that usually played out to be. Just that aspirational, it was marketing and enterprises struggled to truly convert that promise into practice. But the last few years, we've rapidly been approaching a point where, as Tom said, the technology was getting easier and easier. It was also getting more and more capable, which just, it opened the aperture to the number of people within an enterprise that could actually contribute to this digital transformation narrative.
And that's what was so exciting was technology truly was getting . So easy that domain experts and business analysts could turn ideas into actions, and that was worth the time to learn more about it, see what was happening, and as Tom and I found to explore the different realms of citizenry and different ways that technology was enabling this trend.
[00:04:27] Megan Bowers: I think that's exciting that you identified that inflection point that we're actually seeing people be more enabled with tech.
The Rise of Citizen Developers
[00:04:34] Megan Bowers: And I know in the book you talk about citizen developers and other kind of citizen roles, so what do you guys mean when you talk about these citizen roles and why are they important in industry today?
[00:04:48] Ian Barkin: At the highest level, it again means domain experts able to do something with technology. And domain expert means they aren't IT experts. So these aren't folks in your IT dev shop. They aren't hired to be coding in any particular tool, Python or what have you. They're not hired to be specialist data scientists.
They were hired to do their job. In many cases, that job means being an operator in the HR shared service function, or in the finance function or supply chain logistics, or any number of parts of a business that make that business run. I. So what we saw more broadly was, again, a lot of these technologies were getting easier, even to the point, and this happened during the course of us doing the research and writing and then publishing the book to the point of prompting where it was no longer I.
Having to even break down something to a low or supposedly no code framework of dragging and dropping boxes. You could just speak, you could speak your ideas into reality, and that just further pushed that race to that inflection point of. Anybody with an idea can start to use this technology to turn that idea into something, and that something is where the distinctions came.
That's something, an integration between systems and application or some sort of data science model or map or analysis.
[00:06:15] Tom Davenport: Yeah. That ability to speak and get something you want hadn't happened since. I don't know. I got three wishes from a genie earlier in my life and, but we thought that was quite astounding.
And previously people had identified the low-code, no-code movement, which was generally for the sort of small I. Application development and the automation market was developing easier and easier products to use. Ian knows more about that than I do for products that would be user driven. They always tended to have studio in the name, I think, for one reason or another.
And then I, as I say, I was particularly interested in democratizing data science, automated machine learning tools that let non. Professional data scientists do quite serious data analysis and prediction and so on.
[00:07:06] Megan Bowers: I think you kind of touched on this at the end of your answer there, Tom, but I'm curious, like what kinds of applications or analysis are you seeing these citizens developing?
[00:07:18] Tom Davenport: There are a lot, and it's almost every type now, but certainly. Individual level or departmental level applications, keeping track of data for a particular department in a small database kinds of things, particularly as we sell these in oil companies, shell and Chevron and so on. Applications that needed to be developed quickly that it couldn't quite get around to involving disasters.
Say, uh, we're having a big hurricane, we had a lot of damage to our refinery. What are the things that we need to do to fix it back up, get things going again. And if you've talked to it, they, I'd say, sure, I can get to that in six months, but you know, we need to start fixing things tomorrow. So that kind of application and then
Small, relatively small automations typically of a series of defined tasks, not entire business processes generally. And then in data science, it was typically, I would say, non-regulated, non. Critical to life as we know it, sorts of applications involving marketing. For example, if you get a predictive model about who's going to buy something from you a little bit wrong, life is not gonna come to an end
And so a lot of marketing people started to do that kind of analysis and a little bit of supply chain optimization and predicting demand and things along those lines too.
[00:08:50] Ian Barkin: I think one of the interesting things was the spectrum too, because as Tom said, the book stems from two different articles that we were able to publish in two different journals.
One of them had a title of grassroots automation and that idea that individuals in every part of an enterprise have good ideas and have a degree of comfort and familiarity with the scope of work and the job they're doing. This is enabling them to then apply those good ideas and do something with them.
And so we did see great examples departments, like there was an individual in Home Depot who made a huge impact because he had a good idea and was able to act on it, especially in the data science realm. Alteryx came up time and time again as that catalyst for turning those good ideas into those actions and actionable models.
That not only saved hours and hours of analysis, if it was a formal team doing the work, it actually created the reality of the model that wouldn't have happened otherwise.
[00:09:47] Megan Bowers: Definitely. I love those examples. I hear it too, of creating content for Alteryx during the podcast and other things. The ways that people in the line of business can be empowered with Alteryx for sure comes up a lot.
I did wanna touch on something you said, Tom, about kind of the non-mission critical applications and things being put back into the hands of the marketing folks per se. I'm curious what you guys think about in the future, do you think mission critical type analysis will also transfer hands or would that always stay within the it, the data science department.
[00:10:25] Tom Davenport: I think if it's, if something is truly mission critical, then you should probably get some professionals involved, but citizen development is definitely changing the . Process and the participation in those types of projects. Because with citizen development, these subject matter experts, they can engage much more in the process than they could in the past.
They're not just throwing some requirements over the wall to the IT organization and waiting a few months to see something back because they've developed systems on their own. They could even do a prototype and say, yeah, it ought to do something like this. Only it needs to be more careful. Full with critical data or something along those lines.
And Gartner, I hesitate to provide more publicity to Gartner since they do pretty well already. But they coined this term fusion teams, people who are professional IT developers and non-professional business experts who. Can work much more collaboratively than they have in the past, and we saw a fair amount of that at companies.
I still think it's relatively early days for it on mission critical projects, but I think we'll see more and more of it going forward.
[00:11:38] Ian Barkin: It is important to say that we aren't pollyannish or out of touch with the fact that there are risks, and we certainly aren't advocating that you should just unleash powerful technology to absolutely everyone in your industry or in your enterprise and just let them do whatever they want to.
Because that is never a good idea. And actually we focus quite a bit in the book, not only on case studies and the interesting direction of travel of technology and its ease of use and the aperture, more people being able to use it. But we have a framework that speaks to not only the governance you need to put into place, but the guardrails you put in place and the guidance you offer to nurture and educate and enable and coach the folks in your organization that are
Starting to become citizen data scientists and automators and developers. If you're an operator, if you're an executive listening to this podcast, rest assured this is already happening in your enterprise. Your individuals, your citizens within every single nook and cranny of your enterprise are using technology in some way, shape, or form to create solutions for themselves, and it's becoming more and more available.
You can, with copilots built into every enterprise system you've got, now you can start to create these little digital twins and augmentation and applications. So if you think that it isn't happening, you need to rethink that. And then the idea of having governance so that these applications that affect.
Me or a few people or, uh, my team or a department or whole ecosystem, are they edge cases or are they mission critical? All of them will come into the hopper as a result of people being aware of what's possible and coming up with ideas. Then it's up to you putting a strong governance and guard rails structure in place to make sure that those mission critical initiatives aren't killed just because they're mission critical, they're just handled differently.
And, and we have good section of the book that's dedicated to how you might think of doing that. So everyone should buy this book for everyone in their enterprise, make it part of the required reading, and hand it out far and wide.
[00:13:49] Megan Bowers: I love it. We'll definitely link the book in our shout outs post episode for all the listeners to get their hands on it for sure.
But I think your comments around governance make a lot of sense. To me, based on other conversations we've had on the show.
Skills and Traits for Citizen Developers
[00:14:02] Megan Bowers: For listeners out there who maybe are that kind of citizen developer, maybe they're in the supply chain, the finance departments, et cetera, what do you see as like skills that someone needs to be a good citizen developer or citizen data scientist?
[00:14:18] Tom Davenport: I'll let Ian do the sort of characterization of the different types of citizens there are, but there is a pretty broad range of technical skills involved and it depends somewhat on what type of system you're planning to build as a citizen. If it's going to be workflow automation oriented, then you should probably know something about process improvement and how you think about re-engineering a process, maybe even things like Six Sigma and Lean and so on to.
Take unnecessary steps out. If it's data science oriented, you probably ought to know something about statistics, basically, which is the core of that field. And you can generate a lot of quite sophisticated statistical models without knowing much. But to interpret them, I think you really need to have some idea of what's going on.
So I think it depends somewhat and, but as Ian can describe, it also depends on. What type of citizen do you want to be? We actually break it down 'cause
[00:15:20] Ian Barkin: it's a great question and we thought a lot about it and tried to try to come up with a coherent answer. And ultimately there are skills that definitely set you up to be a better and more capable and more prepared citizen of any kind.
But there were also two other elements to that equation, which were personality traits and mindset. So we also focus on that. We can touch on them as much as you want, but some of it was, a lot of these grassroots innovators were seen as rogue mavericks who were going against the grain to even have the idea and to persevere and to pursue it.
And so in many cases, those citizen success cases that. Presented themselves for us to even study. Were a result of people that just had grit and just stuck with it in the face of being told you're not supposed to do that, or are you even allowed to touch those systems, or That's not your job. So there's an element of grit and perseverance and just piracy.
Yeah, that's right. So, so you have to just be bold and so that was a key. As far as the personality trait, the mindset was to have a sort of an entrepreneurial approach to things because you really are, the nature of being a citizen in some cases is doing it for the first time or doing it differently.
Effectively than it is currently being done. And so you have to have that mindset. But then to the skills point, again, it really did come down to some literacies. As Tom touched on. You had to have a level of data literacy and a level of just system literacy and understanding around the processes that happened in your business, which hopefully come as part of the day-to-day nature of the task that you're doing.
But if you didn't embrace and have that, that level or quotient within those areas, then you likely weren't gonna be the right sort of stuff to be a citizen data scientist or automator or developer.
[00:17:15] Tom Davenport: I. There are some skills just to making the connection between what a business is doing and what it might need in the future, and I think in refer to that as the Citizen Scout who can see the need for this, he or she may not be able to actually develop the solution, but they can engage with other people who could do it once that need was identified.
[00:17:39] Ian Barkin: Yeah, I firmly believe because not everyone is going to be a citizen, nor should they be, and no expectation, no training program or initiative within an enterprise should look at everybody and say in in the same way. They couldn't say, everyone is gonna be a black belt and six Sigma. We expect you all to get there.
It just, it wouldn't hold water. It wouldn't be a viable initiative. And so not everyone's gonna be a data scientist, a citizen data scientist, but most certainly everyone given the basic lens through which to look at work can be a citizen scout. I. Can find areas where they say, this doesn't feel as efficient as I thought it might have been before you taught me some things.
Now I know that we can do this differently. Now I know here are areas where it's routine, mundane, transactional, or we could speed up the input of data that I didn't realize was available into models that I didn't think we could build. I'm not the person to do that, but. I know who to call. And so I think a lot of innovation's gonna happen as a result of these scouts out in the front line in the trenches, who now know the art of the possible is there.
They're just not the ones to build it, but they now know who to call.
[00:18:50] Megan Bowers: And having that little bit of data literacy, technology literacy, like just enough to say, I think that there are ways. That we could do this better or I think that this is not the best data method. So, yeah. That's super interesting.
The Impact of Generative AI
[00:19:04] Megan Bowers: I know AI came up a little bit in our conversation earlier around prompting and such, but I'd love to hear what you both find most promising or exciting about ai, generative ai, probably specifically in the context of citizen development.
[00:19:21] Tom Davenport: It's funny, I was, I had a conversation earlier this morning with a venture capitalist friend of mine who heads up. AI oriented VC firm, and she said she believes that basically every SaaS oriented software company will have generative AI as a front end to a user interface. I. To their software and it, Alteryx has started to do that.
It's happening all over the place, and so I think this ability to say what you want and at least get something out of it is going to be just really, really pervasive Now, . Question is just how much expertise do you need to take that something that comes out of your prompts or even to create the right kind of prompts and get an app, a model, a website, an automation, whatever.
And I think it'll vary a lot. And there will be some people who get it over their heads a little bit, but it's bound to . Accelerate the level and speed of digitization that companies can go through. If anybody can create something like that, just by wishing and talking it to existence.
[00:20:34] Ian Barkin: I agree. I'm still trying to get my head around the different spectrum and different elements of AI as it affects this concept and the digitization of work.
I mean, some of it is,
[00:20:45] Tom Davenport: it's also changing every
[00:20:47] Ian Barkin: day, basically , and it doesn't. And so you've got the gearhead geeks who just, who are so excited about just tokens per second and calculations and abilities of models. And I know for an absolute fact, and I would bet. Everything I have on the fact that enterprises haven't figured out a way to change faster as a result of all of this stuff.
I've spent my entire career helping enterprises, and they are very stubborn about making any change, and they're not getting any better.
[00:21:21] Tom Davenport: A friend of mine calls that western one's law technology changes rapidly. Organizations change Slow . Exactly.
[00:21:29] Ian Barkin: And so you might have a fear of being obsolete or missing out, et cetera, but it doesn't change the fact that just your organizational constructs do not allow you to be.
Constantly, you know, experimenting with what has come out in the course of this discussion. Right. There have been 50 new tools that came out since we started talking and they just don't compute at that speed. We better get off and see what they are. That's right. I know. I'm already starting to get nervous.
I'm a little anxious. They gotta catch up. Yeah. If we could take a break for a moment so I can just go see what's happening. But the, there are an expanding catalog of tools that, that help you build applications. And that's interesting, especially for the citizen developer realm, but it still requires that you have ideas.
It still requires that you understand the constructs, and in fact, Tom and I have done some research on the concept of AI agents and what we are now calling Agen X. That's an even more confusing and multifaceted component of generative AI because it speaks to the degree of agency that some of these automations have too.
Act autonomously now. That's a whole other discussion because most enterprise doesn't want agency. It wants compliance with process structure, so, so there's a whole lot going on right now that I think enterprises need to get their heads around. Just because it's possible doesn't mean that it is reasonable.
That said. It's pretty cool what's happening, and you really can just speak ideas into at least really capable prototypes that allow you to advance the discussion with the more technical folks in your organization to turn those prototypes into enterprise, secure, robust applications if they deem them to be valuable.
[00:23:19] Tom Davenport: And with agents, you'll be able to speak your way into
Getting into really big trouble and losing lots of money and so on. It's really quite amazing what's gonna be,
[00:23:29] Ian Barkin: But to be clear possible, but to be clear quickly and efficiently. .
[00:23:33] Tom Davenport: Exactly. With little human intervention.
[00:23:36] Ian Barkin: That's right. So it'll sometimes, I mean, to lose a lot of money over a period of weeks is just grueling, but if you can do it over 15 to 20 minutes, that gets the job done and puts you outta business faster.
[00:23:48] Megan Bowers: Maybe we'll have to have a follow-up episode on agents. It seems like there's a lot to unpack here, .
Governance and Guardrails in AI
[00:23:53] Ian Barkin: Well, and again, back to the sort of the guidance and our governance and guardrails, the reality of any enterprise in any system is that they change, right? That is the only cardinal rules. Everything changes.
I. And so if you do start to deploy an infinite number of agents that work together and on systems, you're welcoming. I mean, and we were talking about gray IT and technical debt with our version of citizenry. Fast forward to the current version of what could happen with all of this ENTs, and it's gonna be chaos if it's not well orchestrated, which is why you should
Turn to the pages that discuss governance, guardrails, and guidance in our book. And one last one, only inspired because Tom had mentioned that most startups and SAEs are going to have a generative front end. There's this other narrative about how software itself is going to. Go away. ERPs will go away because all they do is sit in between you and intent and your data, which is really interesting.
But I think of that as like playing Monopoly without the board. It is just, you need the board to tell you what the steps are. And so logging into no CRM and just talking at your data will take incredible level of sophistication of the people talking to your data. And I just don't see it.
[00:25:17] Tom Davenport: And we need cards that say, do not passcode.
Do not collect $200. Absolutely and a little shoe .
[00:25:26] Megan Bowers: That's super interesting to think about. I've never really thought about that, that piece of it, of like how far it. Could go potentially. But you both mentioned enterprises being stuck in their ways, for lack of a better term, but hard to change and move and implement new technologies.
Real-World Applications and Success Stories
[00:25:42] Megan Bowers: So I'm curious if you have companies that you've worked with, consulted with that are operationalizing AI to get returns on their investment. If you've seen that done well.
[00:25:55] Tom Davenport: I would say absolutely. I mean, you know, one of the big Alteryx users, PWC is all in on generative AI as well, and I'm sure they're using generative AI as a front end to Alteryx, but they, I think that not only for the clients, but internally they'll be able to do all sorts of things with generative ai and they made major investments and most companies are not
All in on generative ai. They're experimenting. They're not carefully testing and measuring how it's working, or they have a sense it might increase. Productivity, but they don't know how much. But, but it is like that. Morgan Stanley is an impressive example. It's not easy to create production applications of generative ai, but they're both doing it and I think we'll see more.
A year ago I did a survey with AWS that suggested only 5% of large organizations had a production application. But now I think in most surveys I see now is between 15 and 20%. So it's, it's growing.
[00:26:57] Ian Barkin: And I think it also begs that you look at what the R is in the ROI calculation and what sort of returns are you looking for?
Because we're coming off a decade in which the overarching narrative towards the end of it was that 75% of digital transformation programs failed. And I. A bunch of these generative AI experiments will also fail. That's just the nature of pushing the envelope and pioneering. But it is changing the vocabulary.
It's changing the way that we look at how we operate and how we understand our data and how we organize as organizations. And one of the interesting. Comments we heard in our interviews for the book was that when you got a sort of a multifunctional team together in a room now, it was very hard to know who was from the business, quote unquote, and who was from the IT side because they were so fluid and comfortable with each other's languages.
And I think as we continue to do these iterations and experiments, we'll just become more and more comfortable speaking to one another, which was half the battle. And most of my career has been spent. I. Evaluating the oil and water nature of business and it not being able to collaborate, those walls are breaking down.
[00:28:08] Tom Davenport: I've spent a huge amount of time on that too. What are we gonna do now, Ian? If those walls are breaking down? That's right. We'll be out of a job. . Yeah,
[00:28:17] Ian Barkin: I guess. Write more books. Just show up on podcasts. Just podcasts not
[00:28:21] Megan Bowers: working on the agents. There. There you go. Make it even smoother.
[00:28:26] Ian Barkin: Honestly, you could spend every waking second just taking YouTube training videos on agents.
There's so much content out there right now. But yeah, Tom, we'll schedule us a separate call to discuss what our career and fates hold for us.
[00:28:39] Megan Bowers: I mean, I'm glad to hear that, that there is butter. Collaboration and speaking of the same languages. 'cause I know so many of our listeners experience that friction working on the data side and talking to the business and vice versa.
So that's,
[00:28:54] Ian Barkin: I mean, the data, the data geeks were the hardest to understand ,
[00:28:57] Megan Bowers: right.
[00:28:57] Ian Barkin: But they knew they were doing something useful. But every group is a specialist group and that's just. The nature of it, and we saw this too with the organization, the evolution of all the digital technologies, robotic process automation, which is where I spent a lot of my time.
Any of the data teams, the ML team, the AI team, whatever they were doing. Now there's gen AI teams, so enterprises tends to create. Siloed center of excellence functions, which are convinced they're the coolest one, and they don't need to interact with the others. And in fact, they should own a lot of what the others are doing.
That just needs to break down, and to some degree it is just because the nature of these tools right now just permeates, it cuts across every one of those sub-functions.
[00:29:40] Tom Davenport: We saw it reflected in the approach that people were taking to Alteryx. I mean, there were some organizations like pwc that saw the benefit of this at a very high level of the organization and gave it all the support that it needed.
But as we said, there were some who just. Discovered the power of Alteryx, you know, and they were resisted by it. At every turn, almost even the communications people wouldn't let us write about some of these people. We, we started out talking about Mr. Citizen, a supply chain expert at a consumer products company who.
Got great benefits out of Alteryx, but had a really tough time getting the IT organization to go along with it. And we were threatened with a lawsuit if we even mentioned the name of the person or the company. So I hope that will, um, make you go and, and wanna read about this. It adds a note of sex appeal to the, to our book, but I think organizations need to realize, as Ian says, this is going to happen whether they support it or not, and they might as well make it effective and.
Get people the tools that they need and saying to them, oh, why do you need Alteryx? You could write that in Python. Uh, excuse me, but I don't wanna write it in Python. I don't know Python. We don't need that kind of language in organizations.
Conclusion and Resources
[00:31:03] Megan Bowers: Well, thank you guys both so much for joining us on the show today.
Really interesting conversation and we'll definitely link resources in our show notes for listeners who wanna dive into these concepts more. So thanks for joining.
[00:31:16] Tom Davenport: Thanks Megan. Nice talking with you.
[00:31:19] Megan Bowers: Thanks for listening. To hear more insights from Tom, you can attend his breakout session at our Inspire conference.
We also have an opportunity for you to download two free chapters of Tom and Ian's book. Head over to our show notes on alteryx.com/podcast for these links and more. See you next time.
This episode was produced by Megan Bowers (@MeganBowers), Mike Cusic (@mikecusic), and Matt Rotundo (@AlteryxMatt). Special thanks to @andyuttley for the theme music track, and @mikecusic for our album artwork.
