Latest / The Tech Matchmaker / Build an AI Agent in Minutes, Not Months? Here’s What It Takes | Shrish Mishra
Transcript
- Gina: Good morning, everybody. Welcome to the podcast. Today we're chatting with Shrish Mishra, who is the principal for enterprise AI and CX for North America at AIONOS. Shrish, welcome to the podcast. Shrish: Thank you, Gina. Lovely to be here. Gina: Now, AIONOS is redefining what a modern BPO can look like. So we're really grateful to have you here. And let's start with the basics and tell us a little bit about yourself. Shrish: Absolutely. It's really refreshing to look at what AIONOS is bringing to the fore here. And thank you so much for having me here, Gina. AIONOS came into being close to about two and a half years back now. And the one problem statement that this company is built to solve for is how enterprise AI has to be delivered. what is the space for outcomes in the enterprise AI agenda? And what I mean by that is that while a lot of AI providers talk about consumption, they talk about model subscriptions and then a license fee and stuff like that. So all of that is aimed at consumption. However, when the enterprise wants to get some real value out of AI, they at times, more often than not, they find themselves alone there. And that's where is the space where AIONOS operates in. So we are an applied AI company and we bring those AI outcomes for the enterprise in five areas, which is customer experience, software experience, data experience, cybersecurity and resilience, and martech and DevOps. The best part is that we completely de-risk the customer from the risk of AI not working because we always come into a relationship on outcomes. So that's the long and short of what operationally, INOS does. Gina: And and you're built as an AI native kind of BPO, right? So blending the live agents with automation from day one. Is that that's correct, right? Yeah. So what does that Shrish: Yes. Gina: actually look like day to day? So maybe we'll take one of those pillars that you spoke about, the customer experience, right? And and who is it built for? And I think the outcome-based is really something that is taking off in the market right now, right? People don't wanna pay for idle time and everything else. And so talk to us a little bit about that framework. Shrish: Sure. So even before coming to AIONOS, I was a big fan of the outcome based model because if you've got a delivery shop which knows how the job is done, it's fairly easy for you to have outcomes in the contract. My CFO does not like me for that, that's been some of the contracts that I've helped author. But with AI, it's only become easier. So when you look at a CX program, there is no, there is no such, I would say prescription as to how much of an AI is good AI. So for us, for a certain business process, maybe a 28 to 30 % AI resolution is good AI resolution. For some other business process, maybe 70 % could be an ideal, AI resolution. So whatever that ideal resolution is, the buck does not stop there for the enterprise. Because once when you handle some part of, let's say, some number of calls for that matter as transactions on AI, there's also a remainder of that population which is still to solve for. So till the time a company or enterprise does not have a unified view of their AI agents and human agents in a single control plane, it's very difficult to manage the life cycle of both AI and human agents. So for you to be able to first arrive at what is the right resolution of AI and then delivering that resolution of AI with the AI agents and the remainder part of it with human agents, yet being able to control it in a single control plane is what we do. And we do it by our orchestration plane, which is called UNIWEAVE. It is a natively homegrown, operator grown product. If you will and in that workspace, you will be able to build agents, run agents, connect your enterprise systems, CRMs, LLMs, SLMs, if you are using any. There is a complete hire to retire mechanism for both AI and human agents. Gina: I love that. And and I think we can both share that, it's it's no secret that traditional BPOs have sometimes have a reputation problem, with the legacy contact centers and the legacy platforms and high turnover, inconsistent quality. What's the biggest misconception that you see today in outsourcing and what do you guys do differently? Shrish: So, I won't call it a misconception, but it's more of a popular belief that... There seems to be an anxiety associated with bringing in AI. You and I were at CCW and we heard the keynotes. So the anxiety about getting AI into the customer facing conversations is real. hear BFSI segment leaders specifically very wary of. at times using the AI part in the customer facing conversations. I was speaking to an insurance company off late and they were not yet ready for AI to touch their customer facing segments. So that anxiety from the operator's lens I find as a real one. And I think what's at the root of that anxiety, Gina, and that's why I saw a misconception, but it's a behavioral aspect that I'm observing. At the root of it is the fact that the inability of the partner owning the end-to-end outcome. So either you've got partners who would say, buy this license, buy this AI system from me and be on your way to solve for the SLAs. Or you would have a partner who would say, you just tell me what is the production number of hours I'll divide it by the FTE definition and stuff for you. There are very few people who are able to say, that we will co-create the ideal AI native operation for you and we'll also own the outcomes in there. So that's what essentially differentiates us from putting our skin in the game and contractually front loading the benefits of AI so that I'm on the hook for the AI to succeed for them. Gina: Which which is like you said, it's a totally different angle for customers that are using BPOs to to experience. Now you mentioned security is one of your pillars or one of the five streams that you have. With cybersecurity, with with these models hacking other models and different things like that. How do you help protect your customers from that? So talk about your security offer. Shrish: So from a cybersecurity, the offering space is one, but even the CX part that we spoke about in terms of UNIWEAVE, all of our products and all of our interventions are SOC 2 type 2 compliant. And when I say we warned the operators had to bring UNIWEAVE and UNISTACK and all of the other portions to the world, it's because of the reason that we are absolutely okay. from that tenancy standpoint, our solutions can be hosted on the customer's premises. It could be on the cloud, it could be on the hybrid cloud, it could be hosted with us. So that's the complete flexibility there. So as far as security is concerned, when we are saying that it will be available in the customer of the enterprise ecosystem, then all the security policies, they don't need to be replicated again. They will be applied to whatever the... infrastructural guardrails are existing within the system. the customers don't really have to reinvent the wheel there. As far as the cybersecurity and resilience offering goes, that's a independent vertical offering that we have for the enterprise. So what it offers is, with our cybersecurity product, which is called UNIPROTECT, what we are able to do is not only secure the the nodes from a cybersecurity standpoint, but the biggest part is the resiliency side of it. So what happens when an enterprise actually receives the threat? What happens when a data breach has happened and you're on the hook for a ransom? And if you could have that one product which says that, okay, leave the ransom aside. Now we have reclaimed all the data and reinforced the security walls against it. Now, even the hacker cannot do anything. So it's in that moment that the resiliency side of UNIPROTECT framework helps the enterprises. That's the whole, I would say, different product. And the last point I want to make is that while cybersecurity as a service is available from AIONOS, at the same time, any product or any service of this AI Applied AI company that we are that the company would use that is a type to SOC to certified and a secure infrastructure Gina: Which is fantastic. If we switch back to the to the BPO side, what does a typical deployment like look like for your company? So when they say, okay, we're gonna go with you guys, how fast can you mobilize a team to be up and running and and get the AI solutions in in place that are required? Shrish: So building an agent onto UNIWEAVE, Gina is not even days, it's minutes. So you can just produce those agents in that control plane. And that's where we've been very vocal about the locus of control to be with the customer. And we wanted that way so that you think of an agent, you want to test it out, you should have the wherewithals without 10 change requests and... a million dollars to be spent to be able to create an agent to test the efficacy of it. So the platform provides for it. So AI agent production is if you've got the SOP, if you've got what that agent has to do, you can build that agent out in minutes, like I showed you in one of the demos as well. Gina: Yeah. Shrish: Now, from an enterprise deployment standpoint, how it normally starts is we start with a joint discovery of sorts, maybe a week in terms of agreeing what success would mean to that enterprise. Why that's important because we an outcome oriented company. We are built by the outcome. So it's important for us to identify what outcomes we are building for. And after having identified and agreed on those outcomes in the first week, it is by the maximum eight week that you can have a full fledged production, which is a hybrid of AI and human agents go live. Gina: Go live. And who's your ideal customer, Shrish? Shrish: My ideal customer profile in the current times, Gina is ranging very widely from SMB to mid market and enterprise. we are of utility to all of these organizations. Largely the ideal customer profile would from a enterprise priority standpoint will be those enterprises who are meaning to de-risk their AI journey. yet want to bear or other, you feed off of the fruit of AI. Because we take away the AI deployment risk from them yet deliver the results of the AI. So that's one behavioral aspect I would look at from a qualification standpoint. The other part is... professionals and enterprises who are meaning to co-innovate. Because when you talk about the call center, so I'll give you an example. The world's largest tax reform is one of the contracts that I led for one of my past employers. So when we were going into that RFP and we're planning with the tax authorities for the world's largest tax reform, prior to that, nobody had... chartered accountants, interning lawyers and accounting technicians as a part of a BPO. They would be considered at different practices as a KPO or something like that. So we provisioned a completely new design of which the delivery orientation was that of a BPO. However, we were solving for the deployment of the world's largest taxation platform. We're taking those queries and we are helping build that platform with that contact center and the back office process that we used to run. Similarly, any enterprise who's wanting to convert their cost center to a revenue center, how would that look like? How can AI play a part there? Gina: Thank you. Shrish: How would those outcomes relay out? Those are co-innovation areas that will be a very strong fit for. And of course, Gina: Two. Shrish: all of the contact center world with high volume traffic where you have some low hanging fruits, low disposition which can be undertaken through AI. Those are attributes of my ideal customer persona. Gina: Do you want to share a specific use case and talk about the added value that you brought to a customer and and what you did and how quick it took them to get live and and what the end results were? Shrish: Sure, sure. So in one of the use cases, the AI experimentation at the customer's end was not really, I would say, I won't say successful, but they had limited success in terms of deploying that. And this is an airline. And also, in another example, it's also one of the GDS platforms, the Global Distribution System platforms. So for us to create an agent, an AI agent on UNIWEAVE, and because UNIWEAVE is a complete platform where it has the context layer, it has the knowledge management, it has multiple layers and two links to which the AI agent has access to. So that platform actually helped them speed up their own innovation to have an AI agent in production for the transaction that they wanted to do just by connecting their discrete data sources, discrete CRMs and SOPs and Excels and whatnot onto this orchestration plane and maybe reduce their FTE requirement by over 30%. Gina: Mm-hmm. Shrish: their time to resolution increased significantly. mean, the time to resolution decreased significantly and they were able to do more with less number of agents. So overall, we were able to, I would say, achieve a 38 % reduction in total cost of ownership for the same process. Gina: It's incredible. As we wrap, is there anything else you'd like the audience to know that I haven't asked you, Shresh? Shrish: No, I think this is one of the most exciting journeys that I am onto. know, off late, Tina, and the conversations with you have been fantastic. I think you are a trusted advisor for a lot of people in the industry. And on the closing thoughts, I would only say that there is a dearth of professionals and enterprises who are meaning to walk the last mile on AI outcomes. But I know stands apart from all that and we start with the outcomes in place. So whether it is about customer experience, when I say BPO, it's not just about the call center, it could be about a revenue program being delivered as a business process services. So co-ideating and co-innovating with the native AI control planes and outcome orientation is what really differentiates us and I'm looking forward to some of the conversations. Gina: Excellent. And to wrap it up, I've always been a fan of James Lipton, so we do a little rapid fire to end the podcast. So what noise or sound do you love? Shrish: I like the rattling of my boat engine. Gina: What noise or sound do you hate? Shrish: the screechy wood rubbing against each other or that screeching sound. Gina: The yeah, it's like chalk nails on a chalkboard. what Shrish: Yeah. Gina: profession other than your own would you like to attempt? Shrish: I love what I do, but I would really want to be a deep sea diver. Gina: Hmm, that's the boat. So Shrish, if somebody wants more information on AIONOS, who should they call? Shrish: They should call Gina and they can call me and I can leave my credentials. Gina: Yeah, we'll put them in the link at the bottom of the show of of the episode so everybody can have that in the in the bottom. But thank you so much for joining us. And the way you guys are rethinking what a modern outsourcing partner can look like is exactly why we keep doing this show. so thank you so much for joining us and I really appreciate your time. Shrish: Thank you, Gina. Lovely to be here with you.