DRUID Talks Ep #4 Is Fast IT a New Industry Trend? with Vargha Moayed

Episode #4 of the DRUID Talks webcast features Vargha Moayed, Business Coach, Advisor and Hyper-generalist, and Subject Matter Expert, Kieran Gilmurray. See the full episode and transcript below.

Kieran Gilmurray: Folks, welcome to our 4th DRUID Talks webcast. Today's topic is Fast IT, a new industry trend for 2023. In a world where everything is on fast forward, we explore how Fast IT can help businesses operate conveniently, efficiently and in a cost-effective manner with the guest of our 4th episode of DRUID Talks, Vargha Moayed. Vargha’s an entrepreneur and advisor to company founders, a venture partner at White Star Capital, a hyper-generalist, and a keen supporter of the latest technologies. And today we're going to cover a range of technologies to help us understand a little bit more about what is Fast IT and who needs it, what and which are considered Fast IT technologies, the benefits and the shortcomings of implementing Fast IT in your organization, the place of conversational AI, which I think, Vargha, you quoted in your article, “Fast IT is part of the dawn of this new technology disruption.” We can't but talk about ChatGPT and its fit amongst Fast IT technologies. And then, of course, for businesses, we're going to look at the short, medium and long-term to-do plan to benefit from Fast IT. Vargha, welcome to our 4th DRUID Talks webcast and thank you so much for coming today.

Vargha Moayed: Thank you for having me.

Kieran Gilmurray: If we kick in, Vargha, with the very first question: What is Fast IT as you defined it in your excellent Medium article - and that's medium.com - are we looking at an old concept or is this something a little bit more recent?

Vargha Moayed: Well, first of all, why did I call it Fast IT? It struck me as an analogy with fast fashion. If you think of fast fashion, what is fast fashion in reality? It's a fashion that changes often or something that is produced very fast every six months or 12 months. It is something that is cost-competitive. It is something that is actually relatively pleasant to use. Usually, they're quite fashionable, with the latest design. And finally, it's a piece that you have in your wardrobe that you might not use forever. It's not a timeless piece. Now, if we make the parallel with the world of IT and information systems, we realize that organizations need fast IT. They need something, an IT that is indeed fast to develop and deploy, that is cost-effective, and that is easy to use. Therefore, you know, pleasant… pleasant, therefore easy to use. I guess that's what I meant to say. And then, indeed, that might not be something that a company might be using for decades, but probably just for months, a couple of months. So that's the parallel. That's what Fast IT is, and I think it is actually a truly new category of enterprise software, you know, and that is underlined by several technologies that we will talk about later.

Kieran Gilmurray: Well, if we jump into that, let's move straight to that if you don't mind. What technologies are part of Fast IT? Because fast fashion, you know, it's something we dispose of, we throw away. It's not meant to be long-term and not meant to be valuable. So, are we suggesting these Fast IT technologies follow that same mold or that same trend?

Vargha Moayed: Yes, in a way. I'm not afraid to say that there is even there a certain parallel. In reality, there are many things. You know, the business world is evolving very fast. Again, in the business plan - I mean, you know, most companies no longer do five-year strategic plans unless you were in the utility, you know, things that require very heavy investments - and in business cycles and innovation have become shorter. As a result, you know, digitization needs, particularly digitization needs on, I would say, back office processes or non-core in the sense of order, support services, processes and in companies who are in the service industry on all of their processes, such as the, you know, banking and insurance, that these processes they change on, or they need to change on average every 18 months or so, believe it or not.

So therefore, they need to be digitized super-fast, and it's okay if they're revisited six months, 12 months, or eight months later, so they are a host of workflows and processes that are not going to stay the same for decades. So, there is no need to be building things that take a year to build because, by the time you build them, user requirements have already changed. So yes, they are, they also have, they can have a short shelf life, and that's okay. It actually exists in companies. I know a lot of traditional IT directors haven't really realized it yet or don't want to see it, but it's a reality.

Kieran Gilmurray: Well, you mentioned a moment ago that Fast IT is needed, and now you've mentioned the Chief Technology Officer or CTO or CIOs; who does need Fast IT? Because in the article as well, you mentioned that the business is a co-producer. So is it just the technology departments that need Fast IT, and those are the people who will use it? Is it the business? Is it everyone?

Vargha Moayed: You know, I think it's everyone. But let me step back a moment and say, indeed, why is Fast IT needed today? Well, let's look at the landscape of companies. You know, companies have been digitizing for decades. And I know we all talk about digitization. It's a very trendy word, but in reality, you know, they started digitizing to such an extent since the minicomputer or so in the fifties. And there are several waves of information technology. And as a result, if you look at larger organizations today, if you look at the landscape of the information system, you know, 30 years ago, you would have few, you know, dozens of applications and few databases.

Fast forward to today; you have hundreds of thousands of applications and hundreds of databases. And they have been traditionally digitizing according to function. So, they started with, you know - I mean, I'm only at the start – the back-office functions with ERPs, front office with CRM. Then they deployed things in the supply chain. And it was logical. Today, because of the digitally native companies that incumbent companies have as competitors and because of the speed in which things evolve, they need actually to be digitizing according to workflows that go from production, wherever that production is, to the customer experience. And they are faced with legacy systems and multiple databases. So, they are not just can’t, like digitally native companies, scrap everything and go out of business for a couple of years until they build something that is according to workflows. So, they need to work with what they have. They need to respond fast.

A 2020 study from McKinsey shows that fast companies outperform their peers by 2.5 times more. So part of being a fast company also means being able to digitize quickly according to the ever-changing demands of your customer and providing a customer experience that looks like a digitally native company’s. So what do you do if you're a CIO? You have to connect legacy systems and legacy databases to be able to digitize according to a workflow. So, for that you have APIs - and I'm not going to get into detail because I want to answer what’s within Fast IT - but therefore, you need to do it fast, so it has to be cost-effective. So, therefore, you need the technology again that deploys fast, that develops fast, it is cost-effective because in IT companies, you know, there is a real problem with talent. There are not enough IT professionals, so you cannot deploy your precious IT professionals to be digitizing things that seem as if they're going to have only a 12–18-month shelf life. That's not your priority because you're still working on the highways of the big infrastructure, and that's what you worry about, and it's not cost-effective.

So, therefore, you need that now. The user, the business technologist - I call them - or the citizen developer, they become co-producers of these needs. So you need a tool that is easy to use, so it allows other people not just to be consumers of it but to be co-producers of IT. So, we can see that there are multiple trends that have come together to make Fast IT a requirement today: Business is going faster, there are legacy systems, you need to be cost-effective, and you don't have enough IT talent. So, all of these are fundamental trends that make the necessity for a new category of tools that I call fast IT, which is a necessity for organizations across the world. And I'm almost going to say, even though I made my example with a very large organization, for across the board.

Kieran Gilmurray: Sure. That's interesting because – again, I think we've seen recently with ChatGPT and other technologies like that – before people thought automation and intelligent tooling and fast IT was very much for blue-collar tasks, administrative, clerical. In your article, you said that “with the emergence of fast IT, there is now a set of technologies that can match the speed and agility of human intelligence, but on steroids.” That's quite a statement and may shock a lot of people who previously thought their intelligence jobs were very much protected. Why did you say that particular phrase… “can match human intelligence on steroids”?

Vargha Moayed: Because, you know, if you look at traditional IT - let's think again, in ERP system. The ERP system imposes its logic on people. It says this is how we built the software; therefore, you need to adapt your processes to ours, and it's very rigid, and you do it in a certain way. And in fact, people realize that – and it’s why Fast IT did it - people create many other things that are not according to the plan because it changes a lot. And that is where we thought that, for the longest time, it was a “pre-be” of humans because only humans had the agility and the intelligence to be able to constantly adapt to changing needs.

Now, when we look at the technologies that constitute fast IT - and let me spell them out a bit now - they are basically things that can use legacy systems and databases to connect applications that exist or to build net new applications using existing databases. So, these are RPA, APIs, large categories of low-code development, and they're all infused with AI, and that's kind of the intelligent part. When you have AI, you can now automate even processes that have decision points that are not only rule-based because, for instance, a robot can, at the decision point, call upon an algorithm to know what path to take. Let's take, for instance, the claims in an insurance company. A robot can put together all the data required for the claim, and, you know, taps into different databases who are the parties to the claim, what kind of policy the person had, what are the circumstances of the claim… And instead of giving that to a claim officer, it feeds that data into the algorithm. The algorithm provides an answer within the confidence interval - and by the way, a human’s confidence interval is 97.5-98. In other words, humans make 2.5% mistakes, usually in their decisions. So, if it comes back with 98% confidence, you should pay the robot continuously.

Now, what have we done? We've done something that is as agile and has used existing databases and existing applications to basically digitize an entire process like several humans would have done. It would have been some humans to enter the data, collect the data, another human becoming the officer to make the decision and another human in the accounting department to wire, let's say, the money to the insurer. So, all of these now have been done with this collection of automation without... and that is why... why do I say that - on steroids? Because obviously, in terms of productivity and the speed at which each works, it is much faster than humans. It's not error-prone. And so, you would still have the agility of humans because it works with existing databases and legacy systems, with the way we do business with a workflow, the way it is. And it can handle exceptions, by the way, because if the AI comes back and says, “Yes, please, you can pay”, but only with an 80% confidence interval, then you can give it to a human which can make a decision. By the way, the decision is refed into the algorithm, which makes the algorithm better.

And then, let's talk about conversational AI. You can even make it easier in the sense that for some of the input and the back put that you have, you don't even have to be typing it any more. You can do it, or you can be typing it, but you don't have to have an interface of use that is a software interface but use human language, which is the way we express ourselves the best.

Kieran Gilmurray: Is conversational AI a fast IT technology?

Vargha Moayed: It is a component of it because, again, when I describe fast IT, one thing that was important was: pleasant to use, therefore easy to use, therefore co-production, therefore easy to adopt. So, a conversation is still the easiest way for humans to provide input and make decisions. So therefore, yes, it is definitely contributing to the family of fast IT because it's a key component of the pleasant and easy to use, therefore, also cost competitive to deploy. Imagine since you were talking about ChatGPT – I don’t know if you want to talk about it, but…

Kieran Gilmurray: Yeah, I have a question for you as well, Vargha... Would you place ChatGPT amongst the fast IT technologies that you mentioned earlier on?

Vargha Moayed: Yeah, absolutely. ChatGPT is a is an enabler. It's an amazing step forward in natural language processing, and so it is an enabler. You actually saw that ChatGPT allows even to do OpenAI Codex, which is basically the ability of AI now to develop on its own code. So, imagine for a minute - and then we are the beginning, you know, ChatGPT is part of what we call Foundational AI, which is models, which is a new category of models. We're basically moving from the artisanal stage of AI to the industrial stage of AI. Four years ago, a model with 210 million parameters like BERT of Google was considered wow! If I'm not mistaken, GPT-3, on which ChatGPT is based, is 175 billion parameters, and the next one is going to be GPT-4, announcing a hundred trillion parameters. We would probably be in two years at 500 trillion parameters. So that allows you to have these models that you can then move - and that's what they have done - they've moved it from conversation to Codex, to image, and it's kind of you just have to tweak them. Anyway, I don't want to get too technical. So, imagine now with Codex and why this ChatGPT, which Codex is just a branch of it in a way, can actually code.

So, imagine you’re a user that is moving up from fast IT to ultrafast IT. You're a user, and you would say to a software robot: “Observe what I'm doing - which RPA can already do today with task mining and process mining - observe what I'm doing and automate it,” and you can say it in language: “Observe what I'm doing now. Start. Finish. Automate it.” And then you already have, you know, again, the “observe” is already done – computer vision, which is the underlying of RPA and then “automate” is basically developing the script, and it can auto-develop the automation script. So now we don't we say even fast IT, it's ultra-fast IT.

Kieran Gilmurray: It's a second article, I think, now, which I'm looking forward to. It's one of the first things I've seen in a while - and I wrote about it recently - that I think is both exciting and scary. It's almost a seminal moment in that ChatGPT has shown that white-collar tasks can be automated. And I don't think white-collar workers were probably living in this false space that suggested that their tasks were so unique and so innately wonderful that automation wasn't going to impact them. Now, I think we're at the start of another revolution that will show us that actually generative AI, and if there are billions and trillions of data points or calculations being involved, now we are starting to match human brainpower. And it'll be interesting to see the next number of years unfold.

Vargha Moayed: Let me ask you two more questions, Vargha if you don't mind. First question: there surely are benefits, as we have described, of fast IT. But there must be shortcomings as well for organizations. What are your key benefits for organizations, and what are the shortcomings that you believe organizations will face? And are there actual ways of addressing those shortcomings?

Vargha Moayed: Sure. I think the benefits we kind of mentioned as we were describing it: it allows to unburden the IT department to be able to actually cost-effectively and fast deploy digitization of workflow, which the companies need to do early because it will allow them to be in this co-production mode and then unburden themselves. It allows having very quick ROIs rather than, you know, again, a traditional IT that was, you know, using years of development. It uses agile methodologies, so it is cost-effective, it is fast, and it solves the talent shortage problem by providing a much faster ROI. Now the challenge, quite frankly, is mainly organizational because, you know, it's new. And you have seen that especially in the case of RPA, which I was involved in for several years, you see that the tools are being adopted directly by businesses and therefore, that quickly becomes a question of who does what the how the work is getting organized between a business who want to develop things on their own and IT that worries about governance, security, scalability, and technological depth, and things that can go wild and unruly. So, the main challenge is an organizational one. It’s to decide who does what, according to which rule. What is the new role of IT from one of doing to one of coaching and providing the overall governance? So, the struggles that we have seen in the early days in many organizations are mainly how to get organized around this. What kind of skillset? For instance, traditionally, in IT, they only had truly IT skillsets. Fast IT, because it's very workflow oriented in addition to truly only IT skillset, also requires a skillset of process re-engineering, a business analyst, and it requires that these people work together. So maybe the question of skills, how do you get organized, what is the overall governance structure, and how do you tackle security issues with this new technology?

Kieran Gilmurray: Fantastic. One more question. What should companies do in the short, medium, and long term, and I’ll let you pick those times, Vargha – if that’s 6, 12, or 18 months to take advantage of Fast IT?

Vargha Moayed: Well, I think step number one is to familiarize oneself with all the different technologies that are composing fast IT and the second step is to let a thousand flower bloom, which is, you know, not being afraid of letting a lot of users develop things of their own and familiarize themselves with these tools and let them play with it. I know this is something that really scares CIOs in the IT department, but it will be a phase like this from which they will reap a lot of benefits. Then there is a phase of putting together, as I said, this governance model on, you know, who does what, according to which rules, what systems are a no-no - that, you know, the business cannot touch. What are the things that they can do? How is the quality control done by the IT department?

So these are kind of the phases: IT to familiarize itself with the tools so that it’s not a mystery and they're not afraid of it, let a thousand flower bloom, encourage actually training of those who are interested in the organization about using these technologies, let them quote on quote play with it and develop their own things. And then on the third phase to say ok, you know, many people have done it, we've got to start to get organized. We are going to have rules of engagement and governance rules, and we are going to clarify who does what, what is done only by IT, is there is quality control by IT or is it only quality control by peer groups, etc., etc.

Kieran Gilmurray: Vargha, thank you so much! You've sold Fast IT to me! If I'm getting efficiency, speed, flexibility, and greater innovation, and we're getting everyone involved in rolling out technology, that is something that's passion in my own heart. I believe, a bit like as you've said and as McKinsey said, faster organizations that can do these things will be a lot more efficient in the digital age than anyone else, and it doesn't guarantee their success, but they're likely to be a lot more successful than those organizations that don't.