← Embracing Digital Transformation
Embracing Digital Transformation
#347 The Future of Workflow: AI, Automation, and Hybrid Work Models
00:34:35
Loading summary
Transcript88 lines
- [00:00]
A
Today, with this as the grounding of how you're going to set up a company operationally, you can do things which you didn't think of before, which means the bar for what profitability growth looks like for every company has changed. What you can do if you are really operationalizing yourself as an agentic company looks very different than the companies who haven't done it that way.
- [00:26]
B
Welcome to Embracing Digital Transformation where we explore how people process, policy and technology drive effective change. This is Dr. Darren, Chief Enterprise architect, educator, author and most importantly, your host on this episode, the future of workflow, AI automation and hybrid work models with Anant Kale, CEO and co founder of Appzen. Welcome to the show.
- [01:00]
A
Thank you. Thank you, Daniel, for having me.
- [01:03]
B
Hey, I mean, we've had to reschedule because I messed up some timing things before, but finally we're on the show, which is great. Thanks for coming. Before we dive into, you know, workflow or workforce automation and things like that, which is our workflow automation, which is a very important topic obviously, and hybrid teams and all that stuff, everyone knows that listens to my show, that I only have superheroes on the show. And so every superhero has a background story. So Nat, what's your background story? What's your origin story?
- [01:39]
A
Oh, as I don't have a very interesting story, but Darren, I grew up in Mumbai in India, did my bachelor's in engineering and did my MBA in finance and that I was working in finance and airtime marketing, a very different industry. I followed my wife actually to the us she was working here. We were married. She was my college sweetheart. And when she came here, I said, okay, we can't live thousands of miles apart. No,
- [02:18]
B
I was just in India. I was just in India two weeks ago. And it's hard to communicate with someone when it's 12 hours difference, 12 and a half hours difference. It's tough.
- [02:29]
A
Yeah, it is hard. It is hard with the time difference. But this was 25 years ago. You can imagine our communication were really, really hard then. Right? Today we can pick up a phone and just call or WhatsApp and see on video. It was really hard then, but it was fantastic. I came here, I started doing consulting, had a really great career, joined in the company called Fujitsu, really massive Japanese congate. Joined as an application architect. My last role was as vice president of applications, where really I had to get to oversee a really massive company. And that was my kind of callings. Living in Silicon Valley and turning 40, I was like, oh my God, what am I doing? Still working. I should be building my own company. That was always a dream. And it just so happened I also met Kunal at the same time. He was my neighbor and we were just chatting on a Halloween trick or treating time and decided this was the right time to partner together and start something. And that's what we started. Absent.
- [03:44]
B
Oh, that's. Well, that's not a boring story. You said it was going to be boring. That's not boring at all. Lots of great things have happened there.
- [03:53]
A
Oh, it's been great. It's been a great journey. I think for everyone who kind of goes into this entrepreneurship journey. Right. There is a very big risk you're taking. We all take that. It's very hard to live this life if you're not ready for what to expect out of it. One of the lessons I got getting into this was, hey, leaving a job, you're not leaving the job because you're going to really make it big on a startup. Because startups, the chances of success are 1%. Right? So if you're going to go in, you better make sure that you're going to enjoy the journey. And that's what I've been living my life since then. It's like, hey, how can I make the most out of it? Enjoy every moment of it and see what happens next. Right? And it's really been awesome.
- [04:40]
B
I love that attitude. I mean, that's a great attitude because yeah, if you're just going into it to say, I just going to do it to make it big, you're going to burn out so fast. Yeah, I love that. Hey, so let's talk about the topic today, which is workflow, automation, hybrid teams, all this stuff. The Internet really introduced it. And I saw it in the 90s, right. The Internet really opened up the. The ability to work with teams that did not sit in the same building together. And that's that concept of that hybrid teams Covid hit. And it re emphasized that all of a sudden for. For a couple years, for a lot of people, we never saw our co workers any. Anywhere down from their chest down. We never saw him. Right. Yeah, right. They were all, you know, who knows if they're wearing pajamas or. Right.
- [05:41]
A
There's a whole new.
- [05:43]
B
Yeah, exactly. So that. That's a big. That's a big. That's a big part of kind of our culture that we have today. I know a lot of CEOs are pushing to move people back into the office with a lot of resistance from a lot of people. So now with AI thrown Into the mix. Now things are even more disruptive. So how do I reconcile all that?
- [06:13]
A
It is interesting you say that you basically talked about. Darren, I feel three really seismic changes that we have had in how we work. I think 2000 Internet and the move to the cloud and being able to connect to anyone, any place, elicipatory of location was massive. I think the second one was really around the mobility. And what happened in 2000, 2007, essentially 8, when the iPhone came out. Right. I feel that was equally disruptive in terms of you're not tied down to a laptop or a machine.
- [06:53]
B
Right.
- [06:53]
A
Versus what can happen through a new interface, which we never imagined could be the focus of every work that we do. So that was second and I think third one, which we happened during COVID really pushed us into thinking about how do we make this virtual thing work? And we probably didn't really have a good answer to it. Right. What you alluded to was a lot of CEOs, including me, pushing for their teams to come back to office is a result of. Yeah, we all went to that stage of having hybrid workforce. And the hybrid was a little different. Hybrid was people in office and people in at home or people remote. Right. Working together. That's what we called as hybrid workforce. How did that work? Because we are forced to do that. And then you could really see it. Some teams working, some formations working, some not working. Or you could see some sizes of. For it or size of companies really not having impacted at all. But certain cultures which are small, agile, where teams are small. Right. Not really being slowing down. Right. So you saw an aspect of hybrid workforce that works for some, doesn't work for some. Today, when you say the word hybrid workforce, Darren, it actually means something different. For me, it's no longer people being in office or in remote. For me, a hybrid workforce today is people versus agents. And how do I think of a team that combines those two together? Right. So that is for me is a definition of a hybrid workforce that has evolved from where you were five years ago to today.
- [08:48]
B
Wow. Yeah. So I never thought of. That's a. We're going to redefine the word hybrid, Right. Because hybrid means a human workforce and maybe an agent work workforce.
- [09:00]
A
That, that is, that is, I think, the reality of what is happening or what you expect to happen in the next couple of years. It's definitely not the commonplace today, but where the industry is moving and where the technology is moving. We are all seeing the impact of AI on the white collar jobs, on the workforce. Right. And when you are eliminating that or you're reducing that, it doesn't get eliminated because you have a fancy chatbot whom you have to talk to, right. That doesn't do the job. It only gets reduced if you have something which is actually a part of your workflow, whatever you're doing. And that workflow step itself goes away because you no longer require that human intelligence, cognitive abilities, which only meant that people had to do that job. And now if that AI, AI agent is able to do it and also execute the workflow, you suddenly have pieces of the work which you were doing day in and day out, not needing people. And suddenly you have to manage a set of agents who are doing that work, but you have to still manage them to make sure they're doing it right. They have, they are doing it correctly and all the other stuff that goes with managing people. So the hybrid workforce concept is, I think, what is really driving the next evolution of the workforce where you have people realistically being replaced or augmented, whichever way you want to call it, by agents.
- [10:28]
B
I like augmented better.
- [10:29]
A
Yeah, yeah, I like augmented too. It feels better than hey, somebody's helping me rather than, you know, taking away my job. But, but, but when you look at the structural way of what companies, large companies are thinking about what they're doing, they suddenly have a workforce which works 24 by 7, which works consistently, which doesn't take holidays. There's no, there's no cost of attrition. Right. They don't have to offshore that jobs because it's too expensive. There's a whole different way of operating a company because of this hybrid workforce. I think that is why it is today. It is so of such a consequence of what is happening around is the shift that is happening around us because of hybrid workforce is probably changing a lot of ways we have done business in the last 20, 30 years.
- [11:16]
B
Well, that brings up an interesting point right there. Because a lot of times we have process or workflow for taking information to give it from one person to another. That's why process a lot of times exists, is how to transform data from one department to another. And you know this, you've been in large corporations, all those transition points are bottlenecks always every single time. That's where miscommunication happens. We need to maybe even rethink how we even see, think about work. Because now an AI can handle the, the miscommunication stuff really well, much better than we can. So, so maybe the work, maybe just automating a workflow. I already have is the wrong thing to do. Maybe it's, I got to relook at, at the whole value system that I have and why I'm doing a workflow in the first place, right?
- [12:14]
A
I think you hit the neon, said Darren. That that's, that's really where the lot of people missed the point. The, the workflows that we were, that we already had design, were designed with the fact that work has to, decisions have to be, have to come from humans. Machines cannot take decisions, right? Because there is some form of data which requires some understanding, some reasoning, and you have the right person to do that reasoning. The right person with the right authority, with the right knowledge. And that essentially what a workflow is, is moving that sometimes in a documented way, sometimes just by communication. Right? But when you start looking at, hey, I designed this workflow in a certain way and I can talk about it more from a finance point of view because I had designed this workflow in a certain way to get to this outcome. But I want to make sure that this outcome is compliant, this outcome is consistent and meets the rules and regulations that are out there. Right? And it's different for different things. You could have a workflow where you're calling a call center to ask help about the software that you're using for recording this. Right? And there is a certain person who's answering it, they cannot do it. And then there is a certain other person who is a workflow that is designed. But if you now think about, hey, my outcome was to help Darren in the problem that he's facing, right? And I have now the superpower of a AI agent who can talk 30 plus languages, who can understand it, who can, who can very quickly go through the logs and figure out what's going on. Right. I don't have to rely on an engineer to do that. The, the help that can offer you and the resolution time I can offer you will be very different than the original workflow that I had.
- [14:06]
B
Well, yeah, of course, yeah.
- [14:09]
A
Right. So that is what is, I think the rethink of the process, which people, which every team doesn't do as good a job as they should be. They try to have, make incremental improvements into the current process because that's probably what we can relate to the most, that, hey, this is my workflow, this is how I do my job. How can I add some layers of AI or whatever, the new things, right?
- [14:31]
B
To automate it, to make it faster, even though it's a bad process.
- [14:35]
A
Yeah, I still benefit from it. But the people who are rethinking that entire process on, hey, I need to get to this outcomes. Right. What is the best way to figure that out? If these are the technologies that I had in place, I think the process looks very different. The problem with that is when you rethink any process to be looking so fundamentally different, the change management around is very hard.
- [15:00]
B
Oh yeah, it's huge. And it increases the risk because processes sometimes were put into place because of something that happened in the company years ago. No one remembers what it was. But hey, now there's this new process because someone maybe got hurt on the factory floor or maybe bad software got produced. So we create all these processes to prevent those things from happening. And there you go. It's there and no one knows why. So when you were talking, when you were. I want to sideline this a little bit because when you were talking, something came to my mind. I have noticed in large corporations, especially you mentioned that, hey, humans need to make decisions. You still need humans to make the decisions.
- [15:58]
A
True.
- [16:00]
B
I want you to tell me when you were at Fujitsu, right? Big company, I met Intel. Now I've been in a lot of big companies. There aren't a lot of decisions being made. People just relegate the decisions through, through the process. Right. They said, well, the process, you know, I'm just following the process.
- [16:19]
A
Yeah.
- [16:19]
B
I'm not making decisions. I'm just following process that I, I think AI is exposing that. What's the right word? That fallacy or that, that issue that we have in companies. Right. That humans are not making decisions anymore. The processes are just being followed. And that's why you have stagnation in these big companies.
- [16:44]
A
No, absolutely. We, we call it different ways. We call it bureaucracy. We, we call it not my job to do it. Somebody else. I'm following the process.
- [16:53]
B
Yeah.
- [16:53]
A
But what it's do is, is the speed of execution and the agility which the company has. Right. So you think about how fast a company can make decisions. The reason. And big companies tend to have far more resources to do that they're creating.
- [17:11]
B
Doesn't that just slow them down?
- [17:13]
A
Yeah, but they have a process in place which slows it down. Where everyone is thinking of how can I make sure that I go into it if something goes wrong. Right. So I have this guy approving it.
- [17:24]
B
That guy. That's it. You get, you hit the nail on the head. Right. I don't want to be responsible.
- [17:29]
A
Yeah.
- [17:29]
B
Right. So I'm just following the process. So that. Right. So. But no decisions are being Made. Which tells me if, if, if we, if we go down your, your reasoning here, if, if AI can do things because it doesn't need to make decisions, it's just transforming data and following a good process. AI can do that all day long. AI can employees that make decisions. I need employees and workers that make decisions because AI can handle everything else.
- [18:01]
A
Yeah, I think there is some reality around it, but the reality of what actually decisions can AI make. Right. So we all know what the limitations of AI are. AI.
- [18:14]
B
Yeah, I can't trust it.
- [18:17]
A
Hallucinate. Right. So the, the narrower you give a scope to AI, the higher the accuracy of what you're going to get.
- [18:25]
B
Yeah, of course, yeah, yeah.
- [18:27]
A
So the narrower the problem is that the data is narrower. Right. So the first part of any decision is how narrow is my AI going to take a decision? Right. So what problem it is solving? Like it could be a, could be a very simple support case that you had opened a question that you had, which is very narrow within the realms of what it is trained on, what it is minded. Right. So you can do much better job than human. And we are finding it all the time when we are comparing human decisions and some of the work that we do in expenses and payables, we find it all the time that the AI took a decision which was actually right when the people didn't follow it for whatever reasons, bias, laziness, not paying attention, whatever the case might be, or not
- [19:11]
B
being able to keep that much information in their head. AIs are good at that, AI is good at it.
- [19:17]
A
But when you take, when there are things which are not clear is when the real human involvement comes in. Because that's the time when you are doing a judgment where sometimes the reasonableness of the judgment is based on your experience, your ability to take a risk. But when you're doing that, you're also armed with all the right information, right analysis, which AI can produce for you very, very quickly. You don't have to wait for days and weeks for analysts to create that. What happened in the past? Let me do the research. Right. That used to take so much time for us to do all that stuff. Not just in my own data. What is happening in the world? How did other companies look at it? All that scraping, putting together that information, analyzing it, giving it, producing it to you, and giving it to you in a way that you can digest and then take your own decision. That this is, I'm. Okay, I'm comfortable with taking that kind of risk is something attributable only to a human. Right. Because we are, we are taking that risk. But collecting all that information to take us there is speeded up 10x because of what AI can do in the small chunks of work.
- [20:22]
B
All right, so I, I love how you positioned all this. Because what that tells me is if I want to keep my job or if I want to have a job in the future, I got to, I got to get stronger on my decision making skills and judgment skills because those are going to become more valuable and based off of my experience and making good decisions and bad decisions. Right. That's how we learn. That's going to make me more valuable in the future. Does that sound right?
- [20:53]
A
It is kind of right, Dan, because that's what you're seeing right now. I'll give you. This is public information around what is happening with computer science graduates. Right. You have seen that the engineering graduates who have graduated, these are kids from top schools. Right. And where just three, four years ago, three years ago, we had a dearth of engineers, computer science engineers. Right. We had to pay premium on the jobs. What you're now finding is people coming out of college don't have jobs, even with these degrees. Because if you look at what's taking away that job is tools like cloud code or Codex or something like that. But what they're really good at is doing some of the jobs or some of the coding that a new engineer would have done, correct?
- [21:41]
B
Yeah.
- [21:42]
A
But they are not able to still match the coding. Eventually they might be, at least today they're not able to match a coding of a person, engineer who has done this for 10, 15, 20 years, who can really optimize it, who can think of security, who can think of all the gotchas that are out there, but they're taking away a junior person's job very quickly.
- [22:03]
B
Well, and that causes a problem because if I don't have junior engineers, how will they ever become senior engineers?
- [22:10]
A
Exactly.
- [22:11]
B
So now I've got a gap. Now I got a gap.
- [22:14]
A
It is creating a big gap. It is. If you're not going to have those engineers who have, who have worked the grind and who have risen up coding it, handwriting, everything, how are you going to get people who are really, really experienced and know to write this, this kind of code? We have, we have a problem. Nobody knows the answer for it. Right.
- [22:33]
B
Well, actually I can tell you what I'm doing because I teach at Vanderbilt, I teach at Vanderbilt University and I teach master's computer science.
- [22:42]
A
Okay, there you go.
- [22:43]
B
So a year ago, I completely scrapped all My classes and I restructured them because before and I in computer science when you take a class and you have assignments that you do, your teacher gives you a test harness, make sure that your application passes this test and gives you all the specs, even write some skeleton code for you. Well, guess what? All that can be generated in seconds by CloudCon cloud code. Thank you. I can't even talk. So a year ago I switched all my classes and now all the assignments that are given are open ended assignments. There's no test cases, there's no code that they get. And the assignments are huge assignments. And they're like, how are we ever going to finish this course? And I go, you have to learn how to use Codex or cloud code. What? And I said, yeah. And you have to justify the architectural decisions that you're making. So I've changed it from a coding class into an architecture class because that's what we need. We need software architects, not software coders. We need software architects that understand coupling, cohesion, scalability, reliability, trade offs between, you know, all the different things. So I switched all my classes to this, this sort of thing because I think that's where the real value is going to be. And, and hopefully these guys can get jobs. Right. When they go and say I've done these things.
- [24:27]
A
No, I'm not, I, this was, I didn't know this like what you have done. I think that is fantastic because basically you are, you are kind of pushing them up the curve.
- [24:37]
B
I'm trying, yes, I'm trying.
- [24:39]
A
You're trying to preserve the curve. Now some people will get it, some people may not take the longer time, but the idea is that you are going to be living with these tools. That's the way to do your job. Right. In the earlier days you had to know how to do Microsoft Office. You were going to be doing an operation shop. You might have to use a tool to write your code. Now you got to figure out how to use an assistant who's going to write the code for you and you're going to oversee it.
- [25:15]
B
Yeah. In fact, I see this weird thing possibly happening that we won't need written human software code anymore, that the geni will write machine code that we can't read. I could see that happening because why? The only reason why we have software coding languages is to give us a way of humanizing the way that a machine operates. Right. And we kept talking about higher and higher level languages. As we move up, the abstractions become more and more and we disconnect ourselves more from the machine. Oh my goodness. What does a large language model do? It understands language. It understands human language. This is amazing. Yeah, right?
- [26:06]
A
What webassaction do you think now?
- [26:09]
B
Yeah, I mean this is totally geeky geeking out. But for example, why do I need to learn and I'm learning this now. I don't need to write my workflows out anymore. I used to code up my workflows in JSON and then I wrote a workflow engine that would take my workflows and do my work for me just so that I could just pass it a JSON file and walk through all my steps and automate my workflows. I don't even need to do that anymore. I can just tell cloud code, hey, I need, I need. This is your data coming in and this is what the output needs to be like and I'm done. So yeah, it's changing a lot of weird and there's a lot of weird things, especially if I got a whole team of agents which is, which are all have their own constraints and they all know what to do. And I mean the workflows can be very ad hoc and fuzzy.
- [27:11]
A
It is interesting. I mean just within our company and where we are managing it, right? We look at it and say that we are telling our customers how to do their finance teams with 90% less people. That's the reality of what our product does. Hey, we put agents in there and but when I think about it, what does it mean for our own company? How can we now support 90% more work without adding any people? That's what is happening. And which means each team has to fundamentally question what their workflow was set up for, how they were doing it. Are all the actors, people, steps required, take them off, what can be replaced with agents or what can agents do and steps can be eliminated. It is a really drastic way of changing. But what it unlocks is obviously unlocks value in terms of cost, right? What it unlocks, but I think what it also unlocks is the, is the speed of execution. That's what I'm excited about the most.
- [28:15]
B
Oh, I like that.
- [28:17]
A
Yeah, even I didn't. I mean today they say the half life of software has gone down even half further down because of how quickly we can produce software. But, but even how fast you can scale a company, the amount of processes workforce that you needed to hire people put that in, even if you had infinite capital would take time. But today with, with, with this as the grounding of how you're going to set up a company operationally, you can do things which you, which you didn't think of before, which means the bar for what profitability growth looks like for every company has changed. What you can do if you are really operationalizing yourself as an agentic company looks very different than the companies who haven't done it that way.
- [29:01]
B
Yeah, yeah, yeah, yeah. So do you think the big companies will be able to adjust to this? Because there's so much inertia, cultural inertia stuck in these big companies that they can't, they can't really move that quick. But a small startup of, you know, three or four people can like, can like completely take over an industry very quickly.
- [29:28]
A
Yeah.
- [29:28]
B
So.
- [29:29]
A
So we are about a 300 people company, right? A little over 300 people. And it takes effort even for us to move. Right. But it is, it is not. We can, we can do that. We have to grind and we can make that happen. Right. It's a. I see big companies to your question because we deal with so many big companies that most of our customers are that. I see two flavors to it. One, there is more often than not most public companies or most large companies, the CEO has made that as an objective. Right.
- [30:00]
B
They're hearing AI. Yes, I've heard.
- [30:03]
A
We must do AI. Right? We must do AI. We should cut our cost. We have to move faster. Right. So they have issued that proclamation. We got to do it now. The middle management, now it comes down to the cfo. CFO wants to do it. I'm going to make budgets available for this. Comes to the cio, comes to the middle management. Right. They are like scratching their heads, what do we do? What does it really mean for me? Exactly? If you go along the path of saying that let me do what I've always done, let me go and get a consultant who will tell me what to do.
- [30:38]
B
Right? That's right.
- [30:39]
A
Yeah. Easy way out. I'll not get fired for it. Right. I got a consultant.
- [30:43]
B
I don't have to make a decision.
- [30:45]
A
I don't have to make a decision. Right.
- [30:46]
B
They're going to give me told me to do it or Deloitte.
- [30:50]
A
Yeah, yeah. And I'll create. They'll charge me a couple of million dollars. Then they'll create a big investment roadmap. We'll go for two, three years. Right. There are companies that are going down that path and then figuring out and have gone down that path. And you have, you have, you have read that reports that came out of McKinsey, right. That 95% of those projects fail. They don't return the R. Yeah, it
- [31:11]
B
came out of mit. Yeah, exactly. So.
- [31:14]
A
So that is the reality of what is happening in many of places. You start throwing money without really figuring out what does it mean for me, the places where I see Darren, where it is successful, maybe those, those 5% where it is successful, where it has still come from the top. You need that top sponsorship because that's the only way to drive a change. But then you are within that middle management. You are identifying within the players, within the operators. I would say right. There are, there are the practitioners, as I would say, people are actually doing the job. You have identified within those people who are the guys who are leaning it, who think of it as an opportunity to grow, lean in, figure out a way of new way of doing work and then empower them to say that hey, take this small area which you are in charge of. How am I going to make that 10x? How I'm going to reduce the cost or speed, whatever the objective is by putting AI and where they become super focused and tight on that and give them a timeline. You cannot have a, you don't have the luxury of going in and hiring a McKinsey to tell you that do it in three months. This has to be come out. Right. I think those are successful teams. At least the ones we have interacted with were very really focused. They know what the ground realities are. They have bought buy in from the people, the practitioners who are going to do it and not just following the narrative from top saying that do this but not really knowing how to make that happen.
- [32:45]
B
All right. So focused, empowered, decision making. These all tie into that success.
- [32:55]
A
Yes.
- [32:55]
B
And that this has been great. We're out of time though.
- [32:59]
A
Okay.
- [33:00]
B
Which really stinks because we could talk for hours. I'm sure on, on this. This has been wonderful. If people want to learn more about you or your company and how to engage, where do they start? Where do they learn? How do they engage with you guys? Yeah.
- [33:15]
A
So we are Abzen Absent is the agent platform for finance automation. We help companies deploy AI agents and save 5 to 7% of their expenses. Reduce the headcount by 80%, make them more compliant. Right. And we do it for some of the largest companies in the world. More than 500 companies across globally use the Appzen platform and we can get it done in a couple of weeks. Right. And you can learn more about it@appzen.com a p p z e n.com and hopefully we have an opportunity to help you as well.
- [33:49]
B
Oh, this is, this is awesome. Nat, thank thank you for coming on the show. We had a great. It was fun. We had a lot of fun. It was fun.
- [33:57]
A
Absolutely. It was great chatting with you, dad.
- [34:02]
B
Thanks for listening to Embracing Digital Transformation. If you enjoyed today's conversation, give us five stars on your favorite podcasting app or on YouTube. It really helps others discover the show. If you want to go deep to dig deeper, join our exclusive community@patreon.com embracingdigital where we share bonus content and you can always connect with other change makers like yourself. You can always find more resources@embracingdigital.org until next time, keep embracing the Digital Transformation.