Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Vijay Tella: It’s Time to Rethink Enterprise Orchestration
176 Plays20 days ago
From Recipes to Reasoning: Rethinking Enterprise Orchestration for AI
“We’re not futzing around with chatbots anymore. We’re bringing AI to the core.” — Vijay Tella
In this episode, Vijay Tella joins HFS CEO Phil Fersht to challenge the outdated patterns of enterprise integration. As generative and agentic AI systems demand richer context and execution capability, Vijay explains why orchestration needs a foundational rethink—and what’s next for the enterprise stack.
What You’ll Hear in 30 Minutes
• Why agentic systems need both rich context and reliable execution
• The mindset shift from “recipe thinking” to goal-based system design
• The rise of enterprise skills as a trust layer for AI
• How to “change the wheels while driving” toward modernization
• Why developers will become essential again—even in a low-code world
Guest Snapshots
Vijay Tella is the CEO and co-founder of Workato, where he pioneered the concept of enterprise orchestration—bringing integration, automation, and AI under one platform. With a background at Oracle and TIBCO, he’s spent his career simplifying complexity in enterprise systems. Today, Vijay is focused on enabling agentic AI by building the trusted infrastructure it needs to act, not just advise.
Timestamps
00:00 — Intro: Why Vijay Tella Thinks Integration Needs Reinvention
02:10 — The Origins of Workato & the “Recipe” Approach
05:52 — Are We Solving the Right Problems with GenAI?
08:45 — The Agentic Burn: AI Hype vs. Real Outcomes
10:30 — Recipe Thinking vs. Agentic Thinking
13:47 — Why Enterprises Must Bring All of Themselves to AI
16:16 — Enterprise Skills: The Trusted Execution Layer
20:33 — Workato’s New Mission: AI That Can Be Trusted
21:01 — The Hidden Integration Challenge No One Sees Coming
23:23 — How to Modernize and Apply AI in Parallel
25:28 — Why Developers Are About to Get Cool Again
Explore More
🌐 Learn more about Workato: https://www.workato.com/
🔗 Follow Vijay: https://www.linkedin.com/in/vijay-tella-1a410a8/
🔗 Follow Phil: https://www.linkedin.com/in/pfersht/
Recommended
Transcript
Introduction to 'From the Horse's Mouth'
00:00:12
Speaker
You're listening to From the Horse's Mouth, intrepid conversations with Phil First. Ready to meet the disruptors who are guiding us to the new great utopia by reshaping our world and pushing past corporate spin for honest conversations about the future impact of current and emerging technologies?
00:00:30
Speaker
Tune in now.
Integrating AI and Data in Business
00:00:37
Speaker
Anyone in the world of integration software has experience working with the platform Wiccato. And today I'm joined by Wiccato's CEO and founder, Vijay Teller, who's lived through several iterations of integration to a world where there is nothing more important than bringing together agentic, gen AI and traditional data to modernize our businesses. So without further ado, let's get the discussion on with Vijay.
00:01:05
Speaker
And welcome to the latest version of the horse's mouth. And joining me today is a well-known face in the world of integration and APIs, et cetera. And his name is Vijay Teller. He's the founder and CEO of Wocato.
00:01:20
Speaker
So welcome, Vijay. Maybe you could give everyone an update on where Wokato is and a little bit about your journey to get where you are.
Challenges of Data Fragmentation
00:01:30
Speaker
Thank you for having me on. It's really good to see you again.
00:01:33
Speaker
So yes, I think, you know, we just finished our 10-year mark last year. You know, you mentioned integration APIs. think the world was always very fragmented. Like there's just a lot of silos of data application.
00:01:44
Speaker
The workflows or processes in the companies are chopped up and all that stuff. But, you know, when we started the company, what we saw was there was just a lot of complexity in how that problem was being solved. You know, in a couple of dimensions, one was there's all these different ideologies, right? Okay, like I'm going to look at it from a data perspective, ETL, like I'm going to load the data into some place. I'm going to do real time.
00:02:04
Speaker
integration. I'm going to do API-led integration. I'm going to like you know maybe like look at it from a process perspective, like a task automation, like RPA, or a you know end-to-end business processes like BPM. There's just an alphabet soup of technologies around how to kind of harness your data applications and processes. And that was too much complexity. And one, the the second part was each of these tools were also individually, like very technical, just specialized people could use it.
00:02:33
Speaker
So when we came into this world with like, you know, the cloud was exploding at that time, 10 years. So, you know, what we said was, first of all, you don't need like six different religions for how you bring systems together.
Wiccato's Innovative Solutions
00:02:43
Speaker
Data, workflows, experiences, processes, APIs, there are different dimensions of the underlying problem that the customers are trying to solve. You're doing employee onboarding, there's a data part of it, there's a workflow part of it, there's an experience part of it.
00:02:58
Speaker
You don't need like multiple tools for all that. So the first innovation was rethinking the world of middleware, if you will, integration. into, it's just one paradigm, we call that a recipe, and the recipe can be like, that handles all of that. can two line recipe that we're moving terabytes of data, 200 line recipe at Broadcom that is like onboarding, you know, like VMware acquisition, and thousands of employees, right?
00:03:21
Speaker
But it's one skill set, one idea that you know unifies all of these different dimensions in one sip one one sort of a concept. But the second thing was make all of this super accessible, not just to like the people that are deeply technical and experts, but also people in the domain, in the business that want to kind of like transform, like, you know, their employee experience, customer experience, whatever, right?
00:03:46
Speaker
So we came at it like a very, mean, deep, you know, to rethink the entire like middleware space perspective, but also make it super accessible. That was our journey. And what we call that approach, enterprise orchestration.
00:03:59
Speaker
And we're not trying to be all things to all people, but, you know, it was really important that best in the world that, connectivity, integration, workflows, automation. Anyway, so that that's sort of the vision that we did.
00:04:11
Speaker
We came into it for the last
Core Business Problems and AI Solutions
00:04:13
Speaker
10 years. And now, over the last couple of years with Gen.ai, I think that it's completely, like, I think changed what all of that means going forward into the future. But just to catch you up on, like, why we started Workado, it was to completely reimagine the broader,
00:04:30
Speaker
enterprise integration, automation, you know orchestration, API space into one cohesive, modern sort of a platform. Do you think we're solving the right problems? I mean, we've done a lot of deep research into the big issues and challenges holding companies back from achieving some of these AI dreams that they're being sold. And the problem isn't so much the tech, it's everything around the technology and especially the process.
00:04:56
Speaker
Then it's the data, then it's the people, then it's the tech. So do you think we're beginning to solve the right problems are we going i run over and over lot of the same insane things like a certain professor once said?
00:05:09
Speaker
I do think it feels very repetitive. And i think what we're doing is we are solving some problems around the edge. What I mean by that is, you know look, I'm going to you know summarize my sales call and like you know you know help with like in ah you know like an email composition.
00:05:27
Speaker
I'm going to automate my account research. I'll deflect some support tickets that are coming into me using AI. I mean, even coding is, you know, it's amazing. Like now, like even the last few weeks, the progress that have been made in that world of what you can do, like you can really replace like a lot of developers with AI.
00:05:45
Speaker
Those are still things that are not transforming your core business of how it works. It's not transformation at the core. It is
AI's Transformative Potential
00:05:54
Speaker
still edge stuff. So I feel like we are just futzing around the edges, trying to, you know, this AI chatbots that they're doing some cognitive things, but there is nothing really autonomous about them.
00:06:06
Speaker
It is, I think, a lot of this edge stuff that we seem to be keep circling around, right? I do think that's what's been going on with AI. Do you think we're going to hit a reckoning pretty soon? Because I think part of the issue has been solving process, which have been designed the same way since the Second World War.
00:06:24
Speaker
companies trying to mask around the cracks with platform upgrades, but not really fun addressing their fundamental legacy debts of technology, of people, of process.
00:06:35
Speaker
So how do we get away from, I'd say, called platform fatigue to more integration necessity? How do clients get ahead of maybe preventing failures and solving them you know and in greater advance? What do you think works?
00:06:52
Speaker
I think, first of all, I do think there's just a lot of really so-so tech, I call it, meaning it just really doesn't really deliver. I think there is a reckoning. I think there is a gentic hype and a gentic burn that's going to come.
00:07:05
Speaker
But I think the underlying potential is serious. So I think, you know, when you look at the platform and how things have been done before, right, even within Workado, I had our all-hands meeting today, and I was talking, ah you know, we're talking about In the first 10 years of even Borchardt's existence, we are proud of like this low-code, hyperspeed, powerful like way of rewiring your workflows inside a company.
00:07:28
Speaker
But even that is predetermined, right? Like we have a problem. you know You want to improve your employee onboarding experience. So somebody is going to like break it down into what are the things I need to do to improve the experience.
00:07:40
Speaker
Then you are building, in our case, recipes. you know like Maybe you're writing code in the case of somebody else. The systems in the history of like technology, from wheel to like a car to like software, mainframes to, you work out of digital native
Agentic vs. Recipe Thinking in AI
00:07:56
Speaker
recipes, they always did exactly what you designed them for.
00:08:00
Speaker
When that's out of date, you have to go back and change it. So the thing that I think is we're missing the boat on is I think fundamentally with AI, it's a discontinuous shift. not just in tech, we aren't doing enough of that discontinuous shift in our thinking.
00:08:14
Speaker
So there's an agentic thinking versus like, I would call it like, you know, machine thinking, which I, you know, I call internally in our world, recipe thinking. Recipe thinking is predetermined top down and you're, know, maybe you do it faster, but agentic thinking is you've got this really like incredible level of machine intelligence that, that you can give it all of the knowledge and expertise, standard operating procedures, you can give it a job description, you can give these things called skills that we talk about.
00:08:42
Speaker
And we'll you know I think there's a role for skills, we can talk about that little bit. And you give it goals, right? You talk about what goals you want to achieve. I want to generate more pipe. I need my you know onboarding process to go smoother and like you know get them productive faster.
00:08:57
Speaker
And agentic thinking is goal-oriented thinking And agentic system design looks different. Tech stack looks different. It's so different. While the tech is here, like from an LLM point of view, and there's all these issues with it and all that, but it's here.
00:09:14
Speaker
i think our thinking hasn't gone through that shift. I think part of the reason we are being fuzzing around the edges is because I think there's some tech gaps there, but there's also like a mindset gap in how we are approaching this.
00:09:28
Speaker
I think you're absolutely right. And I look back to the advent of the internet and why it had such profound impact on business was because it had such a profound impact on the way we use technology personally.
00:09:41
Speaker
What I'm seeing right now, and I talk to senior executives or mid-level executives, kids, is this most people are using ChatGBT claw these tools better and better.
00:09:53
Speaker
And you can tell within five minutes if someone is proficient in these tools or not, or they just sort of be a BSing around the edges. But once you start to, once these tools compile your daily life, you're like, why can't I behave like this my own company?
00:10:07
Speaker
Like I just organized a trip to the UK and I asked, I'm on check GBT pro and I asked it to find me a flight, find me a hotel, rent me a car, find me a restaurant. I just literally talked to the thing and it just was, you know, I'm like, this is getting so good.
00:10:24
Speaker
Like it's getting to the point where this thing's learning me and it's now integrated with my email. my calendar, my HubSpots, everything. And suddenly this thing is doing all the things I've been dreaming of for a while. Like I bought that rabbit device a year ago, which couldn't do any of this because it just wouldn't integrate with anything.
00:10:41
Speaker
But it starts to feel like some of this Gen AI capability is breaking real ground in integration. And how can ah platform like wicado become like that like how can you really identify the way that you operate to make it much more seamless with these emerging tools Yeah.
00:11:03
Speaker
So I think, you know, I think a very simple mental model is like, you know, like, you know, to humanize this, you have this incredible brains, but you need this neuromuscular thing, you know sayinging you know coordination and your hands and things to actually get the work done.
00:11:17
Speaker
Right. I think what has been happening is the brains have just like hyper, they evolved at a hyperspeed, like they're it's so much smarter, like the LLMs are, the power has just been exponential in the last couple of years, but our ability to bring the rest of our company or enterprise to AI so it has all of the context, right? I mean, it will operate with what information it has.
00:11:41
Speaker
You know, it has got a couple of tools, like a couple of APIs. It's going to act on that. Having really complete, current, and correct context is super important or it'll make wrong decisions very confidently.
00:11:53
Speaker
And you also need to be able to take actions in a way that you can trust your core systems with it. So that's where I think our world of enterprise orchestration comes in, right?
00:12:06
Speaker
So when you're connecting to systems, for example, there's a lot of like little technical issues, for example. It's not just API calls. You know, like these systems are giving you updates connecting to Salesforce. It's giving you you, know, new orders, whatever updates. It's not guaranteeing that you're giving you, you know, each of these updates exactly once. so You got to have to process everything that's happening in all these systems, you know, in a transactional way with security.
00:12:31
Speaker
ah So how you connect technically to these things need to be robust. right And it's not just accessing the data, but these are all workflows or processes that the you know the agents are coordinating.
00:12:42
Speaker
You need to have the context on where in process where in these processes you are stuck, right? Like what is you know what's working, what's not. The learning, the context is not just the data level at a process level, right?
00:12:54
Speaker
So you need to have like,
Enterprise Skills for AI Success
00:12:56
Speaker
be fully connected. What I say is in order to bring AI to the core of your company, you have to bring your enterprise to all of your enterprise to AI, number one. And the other part of it is as AI kind of figures out like what is like, you know, and it experiments, it does, it can do A-B tests and comes up with what is the best way to do something.
00:13:16
Speaker
You're going to be taking actions in your systems. And this is something that You need to be able to trust the AI to be able to take actions in your core systems. An important concept that we have arrived at, we we were not thinking about this a year ago when we were starting to work with our first 30 customers on Agentec.
00:13:33
Speaker
But what we found was CIOs, AI is fundamentally stochastic. It gives you different answers each time. And CIOs want predictability, especially when it comes to your core finance and HR and these core business systems, how they work, how you're like,
00:13:49
Speaker
ven a new ah you know when a new order comes in, you create an invoice, that's like a skill that you don't want to reimagine that every single time. You want to know that thing is being done perfectly each time. You know you don't need to like get creative with it, right?
00:14:04
Speaker
And so this concept of skills are going to be important. The human analogy is like, you know, I pick up a glass of water now to drink, right? It's a skill that I learned when I was a baby.
00:14:15
Speaker
When I'm talking to you and drinking the water, my brain is not teaching me from first principles how to pick up the glass of water. That would be disastrous. It would be too much to oversee for a human being if everything is being done right.
00:14:26
Speaker
The fact is it's operating in a set of trusted skills and bringing an intelligence at a higher level. So the thing that we've arrived at working with customers, applying AI at the core of the business, is that you actually need a trusted layer of these enterprise skills.
00:14:42
Speaker
You know how the rubber is going to meet the road with your core systems. You're going to trust that. And you're going to apply the intelligence and like the creativity of AI to come up with new solutions on top of the skills.
00:14:53
Speaker
So there is a kind of a interface between predictability and creativity that that it's AI engineering. We're all figuring this out, right? But that's one of the big learnings that we've come up with, which it's not just about getting all the context, but when you're taking actions to be able to have this foundation of enterprise skills that CIOs would trust so they can take AI and apply them on the top of the skills and they can go to bed at night.
00:15:19
Speaker
Now, AI should be able to create and test new skills as well. You're going to be missing a bunch of skills and there's new ideas that AI comes up with and it needs new skills. But, you know, you ought to be able to create new skills dynamically without somebody coding them up, right?
00:15:33
Speaker
And you got to be able to verify them, test them. So that's where Workado comes in, right? Like we are not like LLM labs. We're not like the creators of AI, of these brains. But we really, i think, have deeply understand what it takes to bring all of the enterprise context and then be able to take, you perform actions, you know, and take actions in a way you can trust. Trust is not just a security and governance thing, which is, of course, the case, but it's also around you know what AI does, you should be able to trust to be able to let it run and do it. And that's the skills concept.
00:16:10
Speaker
So these are things that we are all working through, Phil, you know working with our customers. So it's got a lot of engineering work ahead of, I think, the industry in bringing AI to the core.
00:16:21
Speaker
And we're you know we're starting ah we've started on that journey. But our mission now, I told you about our mission for the first 10 years, our mission now is all around bringing all of that, you know, that orchestration stuff is now just a, that's not the end goal. It's the means to making AI actually work in a trusted way and be able to make really great or even perfect decisions because it has got all of the context, right?
00:16:47
Speaker
Yeah, yeah, exactly. So, As you think along these lines, what do you see as the next integration challenge that enterprises don't see coming? Integration is a kind of a hairy space in a lot of different ways. There's just a complexity.
00:17:04
Speaker
You know, you go look at a traditional company, an insurance company and we're talking to right in the East Coast. They've got data and systems and enterprise context all over the place, right? And from all kinds of legacy systems, SaaS, and, you know, they have data and processes and work happening where the sun doesn't shine.
00:17:23
Speaker
And now you're going to be bringing like AI, you know, you're trying to go agentic. Insurance is like an amazing place for agentic, right? The risk risk assessment is so quantitative, so complex, and There's so much, you know, the climate and all these other changes, they're so dynamic.
00:17:40
Speaker
Being able to do really great, incrementally better and better is like so valuable. Agentic is going to be game-changing, but you're operating an environment where of like all of this really difficult crud, right?
00:17:54
Speaker
So if you have to modernize all of the systems before you actually go AI, I don't think that works either. That's like a five-year project at best, right?
Modernizing Systems with AI
00:18:03
Speaker
So I think the key here is like being able to access and work with data where it is.
00:18:10
Speaker
So you do need to have a modernization approach to your your infrastructure, but you also need to, in parallel, be able to apply with sort of, a you know, kind of a federated kind of a data access sort of a way where you can keep the data where it is and you're still bringing it to the AI while you're also modernizing your system. So you kind parallelize your AI projects with your modernization projects I think that is a you know challenge we see. no I think they we are seeing some significant traditional companies starting on that journey.
00:18:43
Speaker
But yeah, I think it is where the rubber meets the road for AI. and i think you just described how to change the wheels of your car while you're driving.
00:18:53
Speaker
in some ways, even pre-AI, you know, this is what we've been trying to do with companies, right? Like, you know, there's a big Hollywood studio that was lit to the game on going digital, right? So they needed to roll out digital services, but with their existing, like, you know, traditional tech while they're rolling out, like, you know, Snowflake and, you know, Salesforce and all that stuff.
00:19:12
Speaker
In that world, we built recipes that work with their existing systems, like that captured the the business logic of the digital business. And as they were changing out the underlying apps, those recipes like provided like a layer of abstraction that let them do this stuff. So you've got to take an approach where if we can't what we cannot do is to go tell companies, you've got to bring all of your data into one place, you know like into a snowflake before you can actually go do this stuff, right?
00:19:41
Speaker
By the way, this you know this data warehouse data is out of date, the instant you put in there, because the transactional data is actually in your systems. So you got to work with where data and work is happening today while also consolidating and simplifying your tech stack. And you got to be able to separate those two things.
00:19:59
Speaker
Yeah. Yeah. No, I think what you've talked about is a lot of common sense. And, you know, we've been through our own ah gentrification pain. within HFS, we're a research business, and I've seen so many clients go through phase one, which is just finding out how hard this is.
00:20:18
Speaker
But then you're getting ready for phase two, which is, I think, as you described it, which is you have to do some modernization while you're getting used to the AI, you're getting to learn to trust it, and you're figuring out what works. And the underlying issue right now is the technology is really here.
00:20:35
Speaker
There aren't places to hide like there were And we used to talk about RPA back in the day and things like that, right? That's right. Wonderful. I've really enjoyed this conversation. It's taken me back to thinking about all the different issues around integration and how we need to think a bit more about the future and preparing ourselves better and going at it with our eyes and wide open, I think is the conclusion I'd say from just listening to you for last half an hour. So I really appreciated your time today, Vijay.
00:21:02
Speaker
Yeah, no, thank you. Thank you, Phil. I think it's really exciting times.
The Role of Developers in the Age of AI
00:21:06
Speaker
I think one of the things you mentioned, what would be surprising going forward, I would say we'll be surprised how important developers become again. Like, you know, I think we'll we have another conversation at the time, but we were like the you know leaders of this like low-code movement.
00:21:19
Speaker
In a different way, I think developers are going to become important again. And that's my counter perspective on where we're going. MDGA, make developers great again. I love it.
00:21:32
Speaker
Thanks. Yeah, thank you. Yeah, bye.
00:21:38
Speaker
Thanks for tuning in to From the Horse's Mouth, intrepid conversations with Phil First. Remember to follow Phil on LinkedIn and subscribe and like on YouTube, Apple Podcasts, Spotify, or your favorite platform for no-nonsense takes on the intricate dance between technology, business, and ideological systems.
00:21:57
Speaker
Got something to add to the discussion? Let's have it. Drop us a line at fromthehorsesmouth at hfsresearch.com or connect with Phil on LinkedIn.