#294 Bhaskar Roy: How Workato Is Building the Rise of the Agentic Enterprise

0:00Agents are coming in everywhere now. It is not just now swivel sharing between apps, it's also swivel sharing between all of these agents. Even finding a document, you're going through various places. You know, if you want to take certain actions through an agent, you have to figure out, which agent should I talk to? Where is this agent? How do I talk to that agent in order to get things done? Our vision of an agent tech enterprise is that of one super application. Your apps are connected to it, your agents are connected to it. It is bringing in information. It is understanding what is happening in your world.

0:32It's learning through that. Bringing you all the core things that you need to work on that day. And then there are agents behind the scenes powering that work and doing that work for you. In business, they say you can have better, cheaper, or faster. But you only get to pick two. What if you could have all three at the same time? That's exactly what Cohare, Thomson Reuters, and Specialized Bikes have, since they upgraded to the next generation of the cloud, Oracle Cloud Infrastructure. OCI is the blazing fast platform for your infrastructure, database, application development, and AI needs, where you can run any workload in a high availability, consistently high performance environment, and spend less than you would with other clouds.

1:26How is it faster? OCI's block storage gives you more operations per second. Cheaper? OCI costs up to 50 % less for compute, 70 % less for storage, and 80 % less for networking. Better? In test after test, OCI customers report lower latency and higher bandwidth versus other clouds. This is a cloud built for AI and all your biggest workloads. Right now, with zero commitment, try OCI for free. Head to oracle.com slash IonAI. IonAI, all run together, E-Y-E-O-N-A-I. That's oracle.com slash IonAI. Thanks. Thanks, Craig. Firstly, thanks for having me on this podcast. Truly appreciate it. You know, I'm Bhaskar Roy.

2:29I had all inside AI products and solutions here at WorkCardle. I, my career, I started as a software developer way back then. And, you know, this is, I'll age myself by saying this. I was at a company called Tandem Computers, which built nonstop software. And I was a part of the operating system team, building out various kind of capabilities there. And from there, you know, as I started interacting with more and more customers, I loved that customer interaction and then became a product manager, went into product marketing, etc. Now, the way, you know, I came to Workado was actually, there is a story there.

3:13So I started a company called QIK, Q-I-K. We were in mobile video streaming space and then acquired by Skype and then Microsoft. And during that period of time, I met Vijay Thala, who's the CEO of Burkhardt. And a bunch of folks, when we built that team and we went through that particular journey, Once Quick was acquired by Skype, Vijay went in a different direction, looked at something, and then started Workado. And I did another startup in the meantime called Playphone. But, you know, it was pretty much inevitable that, you know, once something happens with Playphone, I'll come back to this particular team because it's a team I enjoy working with, right?

3:59I've worked with them, you know, in the past. So once Playphone got acquired, you know, I came in and joined Workado. And the mission for Workado was what, apart from the people, the mission of Workado was just the thing that just inspired me. Because the entire premise was, you know, how can you go about integrating your applications and orchestrating across the board in a low-code and no-code manner? What I mean by that is, during that period of time, SaaS was growing like crazy. The business team were buying these SaaS applications, but then how do you really integrate? How do you really bring that together?

4:38It was still a very IT-oriented work, right? So they could not scale to handle all the various things that were being thrown at them by the business. So Workado was born to figure out how can we enable not just the IT team to be more productive doing this, but also the operations team in various lines of businesses being able to build their own integrations, the orchestrations, et cetera, and do it in a secure and governed manner. And this opened up the entire aperture as to who all can do things, what can be done. And that was exciting as to how do you go about democratizing this. Since then, obviously, Workado has moved on to the agentic side of things or added, I would say not moved on, added agentic capabilities with that same ethos in mind as to how can we enable people to build various purpose-specific agents that are the core of their business processes and be able to do this in a low-code, no-code manner with all the things that they need in order to deploy agents at core.

5:40Yeah, and this is an enterprise software.

5:46Excuse me.

5:52And I'm sorry, I'll cut all that out. What is, when you say widen the aperture, You're talking about taking it out of the IT department and making it available to people in the business units. And what specifically was the original Workado for? And then we'll talk about Workado Go. Is that the agent building platform? Yeah, so I won't say that it was taken out of the IT department. IT is still the primary user of this, right? It's how do you make them more productive is one part of it. And even when you do that democratization or enable other people to build, IT is still providing the security, the governance capabilities, so that the business teams that are also building these automations can do it in a safe and secure manner, right?

7:01And something that writing can manage in production. So there are two aspects of this. It is democratization, but still with security and governance as the capability there is. So that was the core part of this. Now, what we announced recently, Craig, are two things. One is Workado One as one single platform, and I can get into it in more details. and Workato Go as that end user experience where every employee in a company can become very productive and have just one application to do everything that they need to do in order to get their work done. So those are the two core components that we announced recently.

7:44I see. And Workato, I mean, I always say this when I'm a little bit at sea. If you were to meet somebody at a cocktail party and they know nothing about software without slipping into jargon. How would you describe to them what Workado does? Yeah, so think of Workado as the platform that you use in order to, you know, essentially pull together every aspect of your business and deliver to that. What I mean by that is today your business is run by your business processes. And these processes can be cross-functional or core to what you deliver. How can you really have agents currently as a way for you to orchestrate all of these agents and do that in a secure manner?

8:41And something apart from security, something that scales as your business scales, right? So you don't have to think about all of those core technical capabilities for you to have agents drive operational efficiencies, being able to reduce the risks, you know, overall, and you see ROI from that. Yeah. And it's a platform. It is a platform. Exactly. With different functionalities. and I'm assuming that you ingest data from various places in the enterprise and then makes that data available for different purposes, including agentic workflows. Is that close to what it is? Yeah, it is close. And I can dive a little bit deeper into that, Craig, if you'd like.

9:32I would say that there are five key aspects of this. And this also comes together when you're building agents that drive your core business processes. So one is deep connectivity to your applications, data sources, knowledge bases like you were talking about. Now, these can be in cloud or on-prem, doesn't really matter, wherever your data is. Now, there is an element which is the reason why I call deep connectivity, is it is not just connecting to your applications, right? Whenever you're connecting or trying to move data from one to the other, trying to orchestrate a workflow, there are exceptions that happen.

10:15The end system is down. Part of the data got moved, not the entire thing. So there are all of these exceptions that happen. And the deep connectivity layer that we have takes care of all of that. So that the end user who's building the automation or is building the agent doesn't have to worry about it. So that's item number one. The second is what we call deep action. What that means is when you're going to orchestrate workflows and even agentic workflows, you want that entire orchestration to happen end to end. Sometimes it is connecting multiple systems. Sometimes it's moving data. Sometimes it is being able to understand what that logic is, the reasoning, inferring, all of that stuff so that the agents do a great job end to end.

10:59So that's the second part, which is deep action. The third is agent orchestration and app orchestration. There are multiple apps that you have. There'll be multiple agents that you have. How do you orchestrate all of them together in order to drive that business outcome? That's the third element of this. The fourth, you know, is, as you can imagine, is security and governance. Because all of these things that you're done, that you're doing, has to have, firstly, a zero-trust architecture. But also, if you look at it, when it's going to execute a particular process, I may have different permission than someone else.

11:36How does it take that into consideration when it's orchestrating those processes end-to-end? And then one final, the fifth item, which is in the platform capabilities, there is one single runtime. So you don't have to worry about DevOps. It scales up and down based on demand. It scales as the company scales. It scales throughout as you're building more and more agents or workflow automations throughout your company. All of those things come together. So these are the five, I would say, core components that make up that work out of platform. Yeah. And I mean, you talk about the agentic enterprise.

12:20Can you paint a picture for us? What does the life, a day in the life of a knowledge worker look like in a company that has fully embraced this, the agentic enterprise concept? Currently, if you look at, you know, most of the folks, their day-to-day is swivel chairing across multiple applications. Sometimes I'm looking at a CRM, sometimes a ticketing system, sometimes I'm going into HR because I need to file a PTO. There are all of those various things. And the knowledge worker today is going from app to app to app to get things done. And in future, as agents are coming in everywhere now, as you can see, it is not just now swivel sharing between apps.

13:08It's also swivel sharing between all of these agents. Even finding a document, you're going through various places. If you want to take certain actions through an agent, you have to figure out which agent should I talk to? Where is this agent? How do I talk to that agent in order to get things done? Our vision of an agentic enterprise is that of one super application, which we call Workado Go. Your apps are connected to it. Your agents are connected to it. It is bringing in information. It is understanding what is happening in your world. It's learning through that, bringing you all the core things that you need to work on that day.

13:46And then there are agents behind the scenes powering that work and doing that work for you. So for a knowledge worker, they're spending all the time in this one application in order to get all their jobs done. They're able to search for stuff. They're able to take action. They're able to orchestrate workflows. They even get the insights that, hey, for example, if I want to get a simple thing as laptop, where is it in that particular process? Or if I'm hiring a candidate, where is that agent who is acting on it is in that process? What does that look like? And it's bringing, surfacing those kind of information to me to basically do my job a lot more efficiently and me being more productive.

14:26And so this agentic enterprise, you know, people are talking about this. They're talking about a headless organization. You know, you have an agent that handles maybe sales calls or sorts sales leads and maybe does an initial outreach by email and schedules a call to those that respond favorably. And then you have an agent that prepares the salesperson by doing research on the company and giving him that in a digestible form. And then after the call, maybe you have an agent that's, I don't know what, that's taking the call notes and putting to-do items onto a calendar. All of this stuff, the knowledge worker, he has a web interface with Workado, I'm guessing, and can see all of these different agents and can click on them and see where they are in the process or whether they're waiting for the human to answer something.

15:55Is that, am I describing what it's doing? Yeah, Craig, that is exactly, you know, it's very close to what you're talking about. The other element of this is it's not just the agents. Agents do most of this autonomously, right? But it's also able to guide those agents saying that, okay, you know, I need, help me prepare a call. But these are some of the things that I'm looking for. So those kind of ability for us humans to actually guide these agents so that they do a better job. That's a part of it. And then the agents as a part of, as they're working through that, they're learning through the guidance that they're getting.

16:39Also, for each of these agents, you specify, you know, what is the end goal? What is the KPI? Like you were talking about, you know, the goal is maybe to book a meeting with that particular customer in the initial stage. And so when you specify that goal, the agent works with you to figure out how to go about it. And then it goes autonomously starting doing things. And you're able to see where things are as a result of that. Similarly, after the meeting is booked, like you were saying, a call happened. A transcript comes out. But it's not just the standard transcript. Every company has their own selling methodology, right?

17:14And it's able to adapt that transcript into that selling methodology, automatically updating, say, your CRM like Salesforce and others so that there is very clear visibility or a hygiene of that particular application that comes in. And then when it has to send a sequence, whether it be email or a follow-up call, et cetera, it prompts a human or the sales rep in this case saying, hey, this is what I'm looking to do. Are you good with it? Enable them to edit, make changes, and then it goes ahead and takes the next steps, et cetera. So there is this element of agents and people come together in order to do that work.

17:51And through each of these interactions, agents are learning and understanding more and it's getting better day by day in order to assist this particular person. Yeah. And so you were talking about permission. So if I'm a salesman in an organization, I would have my Workado desktop configured to my domain and my activities. then would my supervisor, maybe the head of the sales department, be able to look across all agents working for all the people in the department? Or, I mean, are there level layers or levels? And then you get up to the CEO and he can look at where Cato go and understand, you know, how far along they are in the sales department to meeting their quarterly target, maybe, or what inventory looks like, if it's a material goods company.

19:00I mean, is it that sort of permissioning so that each person and each level of the organization has a different view? Yes, there are multiple levels of permissioning, right? Based on the role that you have, it understands that. and caters to that role. So at the complete granular level, say, for example, if I'm a sales rep, I have access to certain accounts and opportunities. It understands that. And when I ask to make an update or do some changes, it takes my permission to make those updates, right? So we call that agent authentication. So what that means is there is very clear that I can update only the things that I own or I have access to and not something else that I don't have access to.

19:48So it starts from there, and then it goes up, right? The sales manager is able to see a view of all the sales who report to them. What does that look like? And then as it goes up to the CEO, they see that as well, all the way it travels through. So it respects the role and the permissions that the user has and understands the levels in order to surface the information accordingly. Yeah. And are the agents that are working within the Workado ecosystem, are those built by Workado? Or because increasingly there are these agentic marketplaces where you can go find an agent that does something you want to do rather than building it yourself.

20:34Or is Workado, do you have to build the agents within Workado? That's a great question, Craig. There are two aspects of this. One is we have an agent studio that you can use in order to build custom agents that you want. So that's one aspect of it. The second is we prepackage agents. We call them agent X applications that people can take pretty much right off the shelf, customize it to their specific processes and get going with those. And there is a third element of this. This is what the agent orchestration is, which is if you build agents on some other platform or some other capabilities, our agents are able to interact with those agents in order to perform whatever the function they're looking to do.

21:26So there are these three aspects that enable a company to build agents using Workado specifically, take one of our pre-built agents, or have our agents be interacting with other agents that are built in other platforms in order to get the job done. Yeah. And, you know, there's, I'm not sure it's really happened yet, but people are concerned about agent sprawl, where an organization ends up with, you know, hundreds or thousands of agents. And, you know, no one is really sure how many they are, what they're doing, which is sort of a frightening thing. how do you manage agent sprawl with Workato?

22:18Oh, I think I'll go back a bit and then talk about agent sprawl. I mean, Workato, for the past decade or so, we were helping with app sprawl. There were so many SaaS applications which are there and we are able to connect and get meaning out of it so that you can orchestrate workflows end-to-end. We see the problem very similarly. Agents sprawl, I think, is going to be much more than the apps sprawl. And, you know, being able to orchestrate across those agents is a core thing that we build in. And as a part of that, security and governance, who has access to what agents, what agents can do is a core part, and we provide that capability in order to orchestrate across all of these agents.

23:02And the final answer is, you know, it's one part of it is agent sprawl and being able to orchestrate everything around it. But at the end of the day, if you think about from an end user on employee standpoint, they need one place to go. They don't care if it is agent A, B, C, D, what is getting that work done. They just need one place to go in order to get things done. And that is Workado Go, that one single interface where you go, where you search for things, where you're finding answers for things. And agents, whether it's built on Workado or other agentic platforms, are all a part of it in order to get that thing done.

23:41So the end user experience is also, I think, key in order to drive this adoption. Yeah, that's interesting what you said about SaaS sprawl, because even as an individual, you see an interesting piece of software. You sign up for the trial that is going to turn into a subscription. You tell yourself, well, I'll cancel that before it starts. You try it out, and then you forget about it. And then two years later, you suddenly realize, oh, my God, I've been paying X number of dollars a month for this thing that I forgot I even signed up for. And I'm sure in large organizations that happens a lot. So with Workato, is there a function where you can track all SaaS and agents, agentic applications in the enterprise and see which ones are not being used and then disable them or get rid of them?

24:55if they've just been forgotten about? I mean, is there a level at which maybe an IT manager, a CTO, where somebody can see how efficiently the agents are being used and whether they're agents that people have created and then left idle that should be disconnected from the enterprise. Yeah, there is a, I mean, you can do that with WorkHaro. We call it license optimization, essentially, because one part of it is an application, but typically when you buy an application, you would buy a number of seats, right, in the sense of a number of users who can use that. So, and being able to optimize those licenses based on usage, predominantly based on usage, right?

25:53Things like that. There are capabilities in Workado that enable you to detect that and optimize your licenses across the board. Because we see, especially in IT, we're having to deal with so many different apps and down the road so many different, you know, agents as well. That is a key part as to how do you consolidate that? How do you make sure that you're getting subscriptions for and other things, for things that people use on a regular basis? Yeah, there are certain applications which are not regularly use-based, but that's okay. But being able to optimize licenses across your entire tech stack is a capability that WorkCard provides.

26:34Yeah. And what's the range of agents? I mean, not that agents is the only thing that you guys do, but what is the range of agents that you see in, you know, companies using or building with Workato or that Workato has available in its prepackaged agents? like from what to what? You know, people hear the word agents and they generally think of a very narrow scope or, you know, group of activities. Yeah, so we, in our customer base, we are seeing, you know, a number of different use cases, Craig, but if I look at what are the top four or five, One is the sales part that you talked about. We call our agents genies.

27:35And the reason why it is a genie is typically like you were talking about, agents still are at the periphery of what you can do. They're not at the core of your business processes. Where genies have this capability so that when you build these genies, you give them a KPI as to what is the goal that they need to achieve. Then you provide it with specific skills as to these are the skills that you can execute. So you're enabling that agent to do or its genie to do more. And then you provide it with knowledge bases. And also, you know, what are the various events that it should listen to so that it can run autonomously?

28:14So that makes a genie, right? So we are seeing these genies, you know, in a number of cases. One is an IT help desk. people coming in, asking questions, saying that, hey, my laptop is broken. I need access to XYZ software. And there are agents behind the scene that help you do that. That's one. Very similarly, customer support, right? In the sense when customers are coming to your support site, looking for things, they're interacting with an agent. And this agent is able to, you know, take actions on behalf of this. Like you were talking about inventory. I ordered something. hey, where is my order?

28:51When will I get it? Where is it in that process? All of those things, the actions that agents can take, it is able to do that. So customer support is the second use case that we see. So sales, IT, customer support. The final thing that we are seeing quite a bit is around HR and finance. And HR means because people have to do their job on a regular basis, things like expense reporting, a PTO that I want to file, when should I enroll for benefits? What are the various things that do you on a regular basis on an HR standpoint? That's the other thing we're seeing. And in finance, it ranges across the board, like procure to pay is a pretty big use case that we see.

29:34So there are all our accrual, that's another use case that we So there are all of these various things that we are seeing our customers do. But predominantly, I would say that those are the core functions or lines of businesses where we see where our customers are using Workado agents for. Yeah. Okay. And Workado Go, we started talking about that, is the all-in-one application for employees. Can you tell us about Workato One, which was also recently launched? And within Workato One, as I understand, there's an orchestration. I don't know if it's a level or a module or a component. And then there's Workado Agentic, which introduces the ability to build and manage agents.

30:36And can you sort of explain that and differentiate those and how they relate to Workado Go? Yeah, absolutely. So think of Workado 1 as the platform. You know, we talk about the five key things that are needed in a platform for you to benefit from agents, right? Or bring agents to the core of your business. So Workado is that one platform which brings all of these things together, right? So when you're building agents, say, for example, when a sales ops person is building an agent, they don't have to think of all the intricacies of the platform. How do I run it? How do I deploy it? How secure is it?