Hot Tech Vendors · 11:00 AM to 12:10 PM · Thursday, May 14, 2026
Can everyone hear me okay? I'm just testing the audio. Brilliant. Anyone been here before for one of our Hot Tech Showdowns? That's one. Okay, one in a row. Two. Brilliant. Okay. So I better tell you what the format is a little. We're going to have a series of some of the hottest new emerging businesses pitch you. They're each going to get three minutes. It's a pretty rapid fire kind of play. And at the end of their three minutes, these guys are sharks will ask questions. They get in total about six minutes of questions and then they're off to the stage to the left and they wait until the next guy does their session. Now, the three minutes that they have, they can do what the hell they like with. They might show a video. They might do a little dance. I don't know yet. This is entirely down to them. So I am ready to introduce my sharks. Folks, would you just tell us a little bit about yourselves for anyone who hasn't already seen you? Rohan, do you want to kick off?
Sure, absolutely. I think you've heard enough from me over the course of this morning. My focus is on industry, so I intend to ask these fine folks as to how their value proposition is going to be impacting industries very specifically.
Thank you, Rohan. Hi, everybody. I'm Hansa Iyengar. I'm a practice leader, and my coverage is around enterprise platforms, transformation, modernization, the embedding of AI into it. So I'm going to look at it from a more horizontal perspective, which is like a counter to Rohan's industry-specific view.
Thanks, folks. And I want you to focus really hard on what these guys are telling you because at the end of this, we're going to have a little vote. And the vote is going to be simply you choosing which of these you would back with your own money. So I want you to think about your vote as your own money. It's going to cost you. And I want to see the result of this later. Okay, so with all that said, we're going to kick off with our first, which is Covasant. Come on up. Big round of applause, please.
Yes. Good morning, everyone. Thanks for having me here. There was a video that was supposed to play, but if it's not playing... Sorry. Yeah, so the video just gives you a very... Sorry. Okay.
I have to click here. Oops.
You might have to go two back to make that work.
No. I'm losing time, so... So at Covasant...
There you go.
Thank you. A lot of the AI enterprises want this, but they don't know how to get there. That is where Covasant steps in. We help solve for the AI fragmentation within enterprises. Agents are being built in silos across different workflows, different work streams, across different clouds. Bringing it all together is what we're doing through our Covasant Agent Management Suite. We have multi-agent orchestration that solves for the agent fragmentation. So basically, we help enterprises become agentic by combining CAMs and also leveraging our partner ecosystems and help govern agents and turn AI adoption into real business outcomes. I'll just take a quick example of what we've done for one of the customers. This is a customer that sits, that's a manufacturing customer, but the end customers of this company are retail customers. In this particular case, there were several agents that were built by the manufacturing customer, but they didn't know how to leverage and bring it all together. So what we've done is we built an AI registry that onboards both internal as well as external agents into one place. And the particular problem that we solved here was the frontline retail workers and the associates in this company, for them to answer a query, it would need them to navigate between 15 and 20 different applications, and the latency was very high. So for them to answer a query, it would take 20 seconds. We brought that down to about four seconds. The cost for querying these, the cost for these queries was about $10 for 1,000 queries. We brought that down to about $1.49. We improved the accuracy from 60 to 99% in this particular case. And we are also driving Gemini Enterprise adoption through the particular case study. And because we brought the cost down for querying, this is also outcome-based. So when there are queries at scale, we are now able to bring the token cost down, and we are sharing the cost that has been saved by the customer. And the customer, their end customers are companies like Home Depot, Walmart, Target, and so on. So we are basically an operating layer for the enterprise to drive AI adoption in production. Thank you.
Thank you. And tell the world a little bit about your background.
So I've been a serial entrepreneur. This is my third startup. Prior to this, I built and sold a company, a company called Cigniti Technologies. It was a listed company in India. It was acquired by Coforge a couple of years ago. I saw an opportunity to disrupt the traditional IT services industry. So we started Covasant, which is a Services-as-Software company, the theme that Phil propagated many years ago. We took inspiration from that and started the company to drive services in an agentic manner as well as in an outcome-based manner. Thank you. So over to our sharks.
I'm okay. Thanks, Srikanth. So two industries, specifically healthcare and banking, are struggling to operationalize agentics safely. From a governance standpoint, what have you seen that is working in order to enable that safe operationalization?
So we are a year old, so we are also understanding the evolution of different industries. But if I have to take a case study within the financial services industry, what we have seen is one is risk and compliance based agents. We have seen a lot of traction there. In fact, one of our agents called Kona AI, it is a repeatable and scalable agent bundle, or you can also call it a product because we are pushing it in a SaaS-based model to help identify fraud, waste, and abuse within the enterprises. That's been a great use case. We have 15 Fortune 500 enterprises using this from a risk and compliance standpoint. But the overall theme is still evolving, Rohan. Thanks.
Nice presentation. My question is around when you talk about governance and risk and you're talking about building that layer that connects multiple platforms across the enterprise. What are some of the biggest challenges that you help your customers address when it comes to cross-platform orchestration and governance?
Yeah, so the example is the example that I took as a case study. So this was a manufacturing company. Prior to us coming in, they had eight agents in disparate workflows and using multi-platform, multi-cloud, different ERPs, different CRMs within the organization. So what we had done to unify it was use an agent registry, bring them all into one place, assign role-based agents. For example, in this case, we had a shelving agent, we had a merchandising agent, we had a process agent, we had a retail agent, returns agent, all of them role-based. And then we also have the ability to control the lifecycle of the agent by, if the agent goes rogue, we kill it through the registry, things like that to unified multi-platform agents. Hansa.
All right. I think that's a great example.
So talking about examples, can you give another example of Services-as-Software that has fundamentally shifted operational efficiencies or created a new revenue stream?
So new revenue streams, again, if I am able to take some of the agents that we have built on our platform, some of these agents are repeatable and scalable across different enterprises as well as domains. So that is... So as services, SaaS revenue is not very common, but we have been able to do that through our platform. Secondly, in an outcomes-based model, for example, recently we bid for this proposal for a wealth management firm. So they are having different processes like KYC, onboarding, disbursement, and so on. So all of them, they have about 600 people doing this work for 40 different work streams. Now, we have built 30 agents, along with 15 human in the loop sort of model. And the outcome here is to reduce that team from 600 to about 80. And we are actually sharing the savings after we've built these processes out.
Okay. Thank you. Thank you very much, Srikanth. Big round of applause. I'm used to Silicon Valley pitches and we get whooping and wailing. I need more excitement. If you can just take it to the left, thank you. And our next superstar, KnowledgeLake. Whoop.
All right, hello. Good morning. Thanks, everybody. Oh, this is a clicker? All right. So here we go. Good morning. I'm Russ Malz. I'm the CRO of KnowledgeLake, and I want to thank you for paying attention for our presentation. One of the key things at KnowledgeLake is we've developed a platform, and our platform is around having humans do the work. We are very focused on delivering value for BPOs and GBSs, and we're really focused on heavy document-intensive processes. Last year, we raised $65 million in growth equity, and we're using that money to really invest in our platform and our go-to-market. The document operating system for BPOs really is about three things. One, it's purpose-built. Two, it's AI native. And three, it involves human in the loop. Everything we did was done with customers in industry to make sure our customers could be successful. From our customers, what we're hearing is this is a very powerful platform, it keeps getting better. This is from our good customer, US Imaging. We'll hear more from them. And number two, KnowledgeLake AI works because it augments people, not because it replaces them. Love this quote. This is from an operations manager from one of the teams we work with. And now I want to just introduce you to Darin Williams. He's the executive vice president of US Imaging.
Normally, in a people indexing scenario, we would have about 18 to 20 people in accounting. And now with AI and Verify, we're able to put three people on the, what we call QC or verification of the AI, rather than having 18 people manually index. So the AI indexes, we verify it with three people. Our quality has gone up by 5 to 10 percent, and the workforce has been taken from 18 to 3. As they index these records, they expect anywhere between, you know, two days and 10 days recovery in order to get the index back. We are seeing times of 30 minutes to no more than a half a day. And so the speed is unbelievable. The accuracy that they expect is 100 percent. They haven't been getting 100% from a manual indexing process.
We'll just go from there. So what Darin's talking about is a process we put in place for them, some of the business results that they're seeing. And that's what we're focused on, delivering business results for our customers. We talk a lot about what we want to achieve. We really want to help BPOs and GBSs let the workflow. We want them to really overcome some of the constraints they have, people, process, and technology. So we want to help our customers run smarter, run faster, and run together. That's our promise to the market. When I earned trust, we see really powerful results. Some of them include accuracy, turnaround, records under management, and then a decrease or increase in concurrent projects. So that's what we're about, letting the work flow. And thank you.
Thank you, Russ. My first question for you would be, when it comes to document processing, right, there's a lot of complexity to it. How long does it take for your agents or your technology platforms to get trained up to recognize the kind of document that that particular customer gets? And how does that training continue during the lifetime of your platform's implementation?
Yeah, great question. So we've really learned and really embraced the idea of Services-as-Software. When we set up a job, we bring people. We usually start small. We actually, we always start small. So we look to start with the first job and then scale. Usually you get that first job up, it takes two to four months. It can go faster. It can take less time. But what we're seeing more and more is it's not just getting the software working, right? It's changing the operations. And we're spending a lot of time doing that. As a matter of fact, US Imaging, not only did I sell it, right? I've spent at least four hours a week probably for the last two years with our customers, right? Because it requires that kind of focus to really deliver the results. So we start small, usually two to four months to start, and then we add more jobs. U.S. Imaging currently, last week they had 17 jobs running in their factory with us.
So if we were to pick the manufacturing industry, right, they do a lot of paperwork, operational paperwork, and that's all done manually. Right. So what is changing right now that makes synthetic labor viable?
I don't know if there's one thing that's changing, but I think the promise of our platform is pretty simply this. Right now, some of the constraints that you have, every GBS that I talk to, every BPO that I talk to, they've got multiple systems, they've got highly manual process, they've got a lot of people. If you have a system you can trust that's consistent, that provides single pane of glass, and allows you to run multiple jobs, and know where you are with each job, and process more straight through, you can have fewer people doing more work. You can free your people up to do other things. What Darin talked about, it's really interesting, and it ties to some of the theme of this discussion, was this. With AI and Verify, they were able to launch a new revenue stream, right? So they're doing day-forward processing for counties, which means every new deed that happens in San Bernardino County goes through their system, and within 30 minutes to four hours, it's processed. That used to take weeks and cause some problems for the county.
Thank you. The next question I have for you is like the example, your customers. He mentioned very clearly that they were able to reduce the size of the team because they had to put a platform. Is that something you do like lead? Do you lead with that saying that, hey, we'll help you reduce your workforce or how are you? Where are the people going? Because that's always a big question.
And that's what we've been talking about a lot. Yeah, it's a really good question. So with that particular customer, the original design point was they're going to have to hire 500 to 1,000 people because they had a backlog of 250 jobs. So we were able to avoid hiring. What we're seeing, interestingly, is whereas it only takes now three people instead of 18, they've had to hire certain types of specialists for the U.S. and really transform their QA processing to continue to deliver value. So we don't lead with a cost takeout message. I think that's a really important point. We do lead with a transformation message and the idea that over time we can achieve value.
Okay. Thank you very much. Round of applause for our man. So I'm going to call up Binny now. Binny from Kognitos. Microphone.
Okay, another. Hello. Hey, guys, I'm Binny. And then with me, I have Neerja. She's the chief AI officer. Hi, everyone. So quick question. Show of hands, how many of you can code? Code. All right.
One of us can.
All right. With that GPT. It's OK. Not everybody. I'm a software engineer. Sympathies for all of you who code because software is dead. But for the rest of you who did not raise your hands, maybe in five minutes you'll change your mind. That's our hope. English as code is something that we all humans have been using for millennia. Whether it's the moral conduct code, or it's the legal code, or it's the tax code, or even grandma's recipe for apple pie. These are all ways of programming other humans to do some things in a deterministic manner. What we have done at Kognitos is gone back to this original way of programming intelligence at scale and see can we bring that to AI. And the answer is yes. You can program AI, all of you, in a deterministic way so that you can use it in finance and accounting and other places that are really important. I'll let Neerja talk about a use case.
Yeah, and we have done exactly that for Wipro in the finance and accounting space. And for the international payments for that, which you can imagine actually for a company like Wipro, how complex and the size of those payments would be. The same solution was actually built in bots previously, and it was failing all over the place for a variety of reasons and even deteriorating over the period of time. When we started actually building that as an English as code, we started on day one as an 80% accuracy and stayed through processing on this. But over the period of time, which Binny will go over, with the capturing of tribal knowledge, we were able to actually achieve over 97% of straight-through processing on it. Now, this is not just only the solution. What our forward-deployed engineering team does in the field with Wipro that working across 12 different functions to essentially really identify the opportunities, all the ideation aspect, discovery it, co-build it, deploy it, and manage in the end as well. So it's overall the software and services which we are providing there.
So how does it work? So today, businesses will take an SOP that's written in English and give it to human beings as English instructions. And human beings will process that, and they'll also have some tribal knowledge in their mind, and they are touching these systems of record. This is how most businesses work. What we are seeing in the industry is that we are trying to give that same work, same SOP to Anthropic, Claude, Gemini, all of those things. But AI does not belong to any tribe. So it doesn't have the tribal knowledge that you have with your employees. So it hallucinates. Whenever something is not available, it will just touch your systems. And we are all trying to find out how do we make it deterministic. The right approach, in our opinion, is to create an English as code platform. A platform that can read the SOP, run it as if it was code, not as if it was a prompt. And deterministically runs, if it cannot make a decision, if there's anything ambiguous, go back to the people, ask those people, hey, what do you do in this case? Capture the tribal knowledge, again, as deterministic flow, and continue. We also have an SOP builder that helps you build the...
We're going to have to cut it there. We're on time. So can we go to some questions? Maybe you could expand a little on what you were going to say.
Absolutely, yeah. So in regulated industries, specifically like banking or healthcare, how does deterministic AI change the trust equation compared with automation? And if you can give us some specific examples, that will be helpful.
Yeah, so just like automation, deterministic AI will not hallucinate. However, automation was brittle. Bots would crash and burn when any one piece was missing from the automation workflow diagram. In our case, it pauses. It reaches out using an LLM to a human and explaining what happened in the deterministic flow. And then with the human's help, it'll change the plan, continue, record that as a new knowledge article, so that next time it can be smarter.
And in addition to that, it's also giving you the auditability on it. So everything which Binny has described, there is actually in a platform, the auditability and the governance is the core part of it. Every single English code which you have really built and executed, you could audit even one year from now.
So when you take instructions in English language and you try to convert that into code, a lot of emphasis goes on being able to articulate the requirements really, really clearly. Because the more ambiguous you are, you take so many iterations and it's going to break. So what are you doing with your customers to help them, or rather with the users of your to help them be better articulators of the requirements?
So we have an SOP builder. So the SOP is written by AI. So our builder talks to the business person and says, do you have a workflow diagram? Do you have anything? Or just give me an example. And by talking to the business user, it writes the detailed SOP in a language that is detailed enough for our native English interpreter to run it. So we're not actually translating it to yet another computer language and hallucination somewhere in between. It runs this detailed SOP natively.
Okay, and kind of a follow-up question. How much of, let's say, coding or technical knowledge and skills does the user actually need? Because you're now giving the power to, taking the power away from an IT guy and giving to a business user to be able to develop in natural English, so how much —
That's what I said. Everybody, even a teenager, should be able to figure out how to program in English. If you ask somebody, hey, where is Starbucks? They're programming you. Go there, turn left, turn right. Everybody is a native programmer. We just haven't been told that.
All right. Rohan, do you have a follow-up?
We have time, yeah. We do have time for one more question. So if we just think about manufacturing or supply chain, a lot of their operational processes are based on tribal knowledge. It's not documented. So you talk a lot about English as code. How do you approach preserving and operationalizing that tribal knowledge?
Yeah, yeah.
So we have actually done that with a couple of manufacturing already. The way this is really done is that once the process is really built, as Binny was explaining in English, whenever you hit an exception in that, right? So for one of the manufacturing, they actually do the whole project-based systems, which is essentially building the interior design for the likes of McDonald's and Starbucks, right? So their process is so ad hoc in nature, right? So essentially, there are a lot of exception scenarios on that. So we have done on their sales cycle, on their project management, on their estimation part, on their lots of reconciliation also from BOL to whatnot, right? So all those steps which you are really talking about are really captured in English. And once at one time, they don't have to redo the same thing next time.
Yeah, think about it as continual learning. So anytime you hit an exception, our English interpreter pauses and involves a model that is trained to understand the change in the business process that is needed to proceed with that use case, helps solve that one problem, and then authors a knowledge article, again in English, for review by human later on, so that somebody can say, oh, I like this one use case that was done. And I would like to use it without human involvement in the future, or maybe say, go back to the employee and say, why did you approve that? It should not have been approved.
Okay, and what about the security and governance piece when it comes to having users code whatever they feel like coding?
I'm afraid your answer will have to be quite quick. Yes, no, we are not trying to change anything in businesses, the way you operate. Who writes the SOP in your business? Not everybody. So the same person is going to write this thing, right? So the governance should not change. The way you govern humans at scale, you should govern AI at scale.
All right. Thank you. Thank you very much. Thank you. Thank you guys. A round of applause for Kognitos. Thank you, guys. If you can exit that way. So next up we have Lyzr. Now, correct me, is it Lyzr or Lyzr?
I'm a Lyzr.
Lovely. Okay, I finally got it, Lyzr. Come on up.
Take a mic. All right. Good morning. So today, I'm going to speak about Architect, one of our newest innovations that we launched. And this is the URL, by the way, you could try. The biggest bottleneck in enterprise AI is actually figuring out what to automate. Where can AI give them the highest ROI? So what do enterprises do? They bring in a consultant, an external consultant, who can speak to their business teams and come up with a plan after 12 months. Or they'll get all the leaders into a boardroom and figure out what to automate. But we've seen that both these have been very ineffective, be it the U.S. government, our largest customer, or be it the numerous Fortune finders that we work with. In fact, your business users know more about your business than McKinsey does. So what if we tap into that? This is a real scenario. One of the largest banks took eight and a half months to ship one production agent. Why? PRD took a month, build took four months, and the feedback took three more months to get the agents to production. Because the people who know the problem were asked to wait, the business users, and the people who don't know the problem were asked to build. This is not just a tooling problem. It is a gap in this whole, it's a wrong loop that we're looking at. So we built Architect. All you can do is just go in, share your problem or share the agents that you want to build. It will look into its blueprints, thousands of blueprints that it has. It then builds out the plan first, clearly choosing the agents, choosing the models automatically. It also writes the UI UX spec for what you're trying to build. It then orchestrates the entire agentic system right in front of you. Click, you can see the prompt, it writes the prompt, it selects the models, temperature control. You don't have to move a needle here. And then it also builds an app for you. Because the best way to interact with agent is not via custom GPT. It is via an actual app. Because for the last 20 years, we've seen the app interfaces, and we react to apps much better. That's how the human brain is wired. But the best part is, while your business users are building these agents, all the agents automatically get registered into the central agent registry behind the scenes with all the CIO controls that you have in place, like responsibly guardrail, security policies. So the CIO can actually push the innovation to your teams without having to worry about security. None of the agents go to production. They get into simulation, 10,000 simulations per agent. It runs the simulation. It checks where the agent fails, and through a reinforcement learning loop, it fixes it automatically. Imagine 10,000 simulations. It takes four months for a human to do it. It does it in 40 minutes, and then you go into CI/CD pipeline, which is very important. We're the first company to launch agent CI/CD pipeline. And now it can get all the approvals from CISO office, from every other team. And finally, you get a control plane. You get to see all the agents that are in production. What are they doing? Is there an agent drift? But behind the scenes, Architect also tells you what your business users are building. It does cluster analysis. It also does impact analysis to tell you, hey, this is what your teams are struggling with, document processing and so on. It actually builds you the agentic pipeline. It tells you the ROI as well that you could get. So that is what we built in use at several Fortune 500s. You are literally crowdsourcing your agentic pipeline, not asking McKinsey to come in and build. So try out architect.new. I built a math tutor for my five-year-old, so it always works. Thank you.
Thank you. Great presentation. So I want to understand when your AI is architecting the agentic workflow, how does it account for where humans need to be in the loop and at what level do you kind of have some actual person touch the code and touch the tool you're building?
Fantastic. We are, again, the first in the world to launch SuperFlow, which is the most complex orchestration. I mean, it can handle a lot more complex orchestration, 10 degrees above what N8N could handle. So the agent understands where to bring in humans in terms of, because agents always are handling judgments, not deterministic outputs. And we also, in the interface side, we have a concept called agent decision inbox. When the agent is unsure about an output, it brings to the decision inbox, where as a human, you can check and approve or disapprove, so the agent learns from your feedback. And this goes in on a loop, right? So I can easily say that in a few years from now, your workday would look like an email inbox and an agent decision inbox. Agents would have run overnight, and they'll be waiting for you to come in and approve the actions. So agent decision inbox is the human interface side of things, but behind the scenes, agents know the human in the loop to bring in. That's because of the blueprints. We have over 1,000 blueprints that we have deployed for our customers, and those blueprints allow the architect to decide where to bring the human in the loop.
So you probably are well aware that banks and financial services really want to scale up AI agents.
Yeah.
So what would you say are the three biggest barriers to doing that, and what are you doing to help overcome that?
Yeah, the first one, like you saw, was figuring out where to start because they always start with customer support, et cetera, which is okay. I mean, there are Sierra and multiple other companies doing that, but the real ROI comes from a lot of things that you won't know, document processing, or it could be, say, loan origination systems, et cetera. So first problem is this. Second is productionization. Once they build the agent, how do I know that it's going to work when I'm sleeping? So that's why we build the agent simulation engine under the CI/CD pipeline, so that it can go through the simulation, get all the approvals in place, and you go live. The third one is what we call as agent drift. Because the models change behind the scenes, you saw Claude dropping its capability all of a sudden. Agents don't know. We humans know agents don't know. All of a sudden agents will approve a claim which it's not supposed to do. So that's what we call as agent drift, actually. So we also built agent improvement engine to handle that. These are three things that we keep hearing from our customers, banks, insurance firms, because the cost that they pay is massive when something goes wrong. I asked one of the largest banks, I cannot name them, but I asked them what they considered their kind of failure in the last 12 months. They said, we were not able to ship an agent that our customers can use to check their balance. We have very deterministic chatbots, but not a conversational agent. Because what if the agent says you have $5 million in the bank? What if the agent says that to 10,000 users? Everyone is going to take a screenshot and start suing you. And that is where the banks and insurance firms are stuck because the cost they'll pay is huge. So these are the three things that we keep hearing all over again.
I think we still have time for one more question. So when it comes to actually integrating the agents that are being created in a multi-platform environment, which becomes very difficult, very complicated, how easy or how hard is it, and how do you kind of help your customers integrate?
Integrations are actually, to be honest, have gotten a lot easier. You have API, MCP, CLI, multiple ways to do. But you don't have systems. You have human API. So as a human, I'll download this sheet and I'll upload it here. So those things are there. And we see that in the U.S. government and a few other scenarios. The issue really is what is the agent authorized to do? So we've designed our system in a way that the agent always takes the user's credentials. If you have access to a certain table in Snowflake or a certain SharePoint folder, it'll always just stick to that, basically. So we follow the user authentication. And you can also build agents which has the central admin auth as well so that it can literally do a lot more work. But then that poses a different problem. An agent that has admin controls or admin access, what if another agent tricks it via prompt injection and take the data away? So this is, again, another large bank asked us, how do we solve this? And they actually worked with us to solve this. We built something called agent entailment policy. Similar to how we have human entailments, now agents have entailments. It has a central policy that tells the agent what to trust, what not to trust. And you can also set up policy in a way that do not trust anyone, literally anyone. So those are checks and balances. And to be honest, we are still scratching the surface. We are still in year one of next 100 years to come in this space. But yeah, so far we've gotten far this much. And I think we'll continue to just listen to customers, understand their pain points, and add more features to the platform. Thank you. Thank you so much.
Thank you. Thanks very much. OK, so our final act is OrbitShift. And Saurabh Mishra is coming to join us. Hi, Saurabh.
Thank you, everyone. So I'm the last man standing. So just a small caveat before I begin. This presentation might sound a bit too simplistic because our audience is the sales leaders, enterprises, leaders, and teams. And last three years since we have been in existence, all our hypotheses on their tech savviness have been shattered. So we keep things very simple, whether it's a presentation or our platform itself. Sorry. This is nothing derogatory to anyone in this room. OrbitShift is a system of work for the enterprise sales team, as I mentioned. And this is a cliche now, word is going through strange times, and enterprise sales, sellers, sales teams, leaders, all of them are actually navigating pretty ironical times, if I may say. Selling is becoming actually tougher and tougher. Salespeople need to cover a lot more accounts to actually manage the quota. But if you look at the level of details or the preparation they need now to really win the accounts, it's gone actually a lot more. So more accounts, but you need to work harder in each of those accounts. Second, as you cover more account, you basically need fewer incredible sources of the data, information, intelligence. But I don't know how many of you know, but in the last 24 months, the number of products in the sales, the launches, have been more than collectively in the last 10 years. A decade. It's mind-boggling, actually. Right? So the confusion glows. And the final thing, the most important thing, is every organization has a different offering, different product services, you know, and the coverage bar going on higher and higher. You need a lot more contextual stuff. You need, you know, a lot more customized stuff, relevant stuff. And instead what's happening is that there is like a plethora of the LLM wrappers, right, which are scraping the web and feeding the general information to the users, basically creating confusion for them. And therefore, our proposition to the our proposition to our clients is very simple. There are three things which we stand for. One, this is a full sales cycle enablement. Right from the early stage of identifying accounts, what matters there, lead generation, to the deal shaping, all the way going to the proposal management, full platform, full sales process orchestration gets done on the platform itself. Second, this is purposeful for an industry. I mean, we strongly believe that AI adds more value, especially in the enterprise sales, when it talks to, if I'm in a cybersecurity company, then it needs to talk to my requirements, technology services providers, it should talk to the technology services provider context. So all our models are custom trained for a particular industry we serve. And third, there's a highly curated data lake. Our platform doesn't actually scrape the web at all, right? It actually curates the, you know, the data from hundreds of partners through our own data lake, right? And what it means is basically a few very tangible things for our clients. 60 to 80% reduction in number of existing sales solution databases. So this becomes the core engagement platform. Zero ramp up time because it's purposeful for industry. I'm selling, I'm a service provider, for example. I don't need, you know, I don't need even training actually on the platform, literally. Because it's just everything is so intuitive, purposeful for me. And third, because of the curated data lake, everything is fact based. There is no hallucinations. So that's the, you know, a bit about the proposition. And that's it. The final thing is, it's not about enablement, actually. It's about the outcome delivery. And that's our simple — salespeople know that very well. So we don't tom-tom enablement, we tom-tom the outcome delivery in terms of the pipeline and deal wins. Okay, OrbitShift.
Thanks, Saurabh. You sort of described the problem of abundance in the world of sales. We also recognize that salespeople are generally overwhelmed with the number of the fragmented data and the number of AI tools out there. So what specific problem are you trying to solve for differently? And if you can give us an example in a specific industry, that will be really helpful.
So we serve technology providers quite deeply. We serve actually eight out of the global top 15 providers like Infosys, HCLTech, et cetera. I'll give you a real example of a salesperson. You know, here, if I'm selling, for example, I'm just making it up, selling to JPMorgan Chase, what will I do? I will actually look at any product or any, it's a ChatGPT or any other sales product, I will actually go and ask what are the priorities of JPMorgan Chase, what can I sell them, here is what I have, tell me who are the relevant people. OrbitShift actually cuts all of that. What it does is the platform is already trained that, for example, as a service provider, it will take in here are the 10 offerings, right? Could be product services solutions. This is what you sell. It will scan JPMorgan Chase, you know, all the data. And it will tell you that out of your 10 things, here are three things which matter at this time. Here are the five people who are probably the decision maker. Here is what is their agenda and this is how you should pitch. And here is your sales pitch. So it's just, you know, just cuts through all the noise on my day-to-day work, right? I come here and I know what to do. So it's about the actioning, not even actually about the intelligence, because that's the background work.
Thanks. Can you talk a little bit about the technology under the hood and how that's making all of this possible?
Yeah, absolutely. So look, right from the start, we took a call that we are not going to be a data company because, you know, in the AI era, data is in abundance. And therefore, our approach was very simple that we're going to buy the data. And for our enterprise clients, all the cybersecurity, all the security, infrastructure requirements are quite, so we can't scrape the web in general, right? There's a high bar. So we have multiple partners, we buy the data, and then we have created our own, you know, so we deploy a bunch of LLMs, which are custom trained for the different types of work. So for example, if you're IT providers, there is a BPO, engineering services, or enterprise software category. So we have different LLMs which are trained on their context. So each of those industry, we look at thousands of the company, what kind of offerings are happening, all the embeddings, et cetera, are trained. And then, you know, once we serve a client, basically we set up them the very contextual setup for them. And there are multiple LLMs we use, you know, for description purposes, for the reasoning, a bunch of process.
So my follow-on question is that the examples of clients that you mentioned are very specific to technology companies. How would an enterprise like maybe a manufacturer or a bank use your database and your tool?
Absolutely great question. And that's sort of in line with our thinking that we will customize the models to the industry we serve. So we are expanding, for example, we've now grown to the broader professional services segment like staffing services agencies, et cetera. We've started actually serving some of the clients. Progressively, we're expanding the ambit. At this point of time, there are a few areas like the medical devices, et cetera, where we have sort of already engaged with the clients. We have now understood how to build the contextual layer. Therefore, for us, it's now a couple of weeks. It's almost now a modular structure for us, right? So if a new industry comes in, in a matter of a couple of weeks, we can frankly create a completely new instance for those guys.
Wonderful. Thank you. Thank you so much. Thank you, everyone.
Thank you. Okay, so I don't know whether I need to do the clicking, but we need to get onto the vote page if that's possible. I'll try clicking myself. Yes, so we are now going to have the poll. I'm hoping that you all have your apps available to vote, and I want you to vote as if this is your own money that you are going to invest in one of these firms. Now, they're all going to get a prize. They have all great hot techs. They've all been already selected by HFS for that. So don't feel like you're doing any harm at all. You're just positively showing where you might want to invest. Is that poll running now? We can show it on the screen. Thank you. In the meantime, folks, our sharks stop staring at that. Sorry. What did you make of the presentations you've had today? So we've had five pretty interesting and quite different propositions but they're all very technology driven. Is it what you were expecting from this or would you expect something even more shocking and different and should we vote accordingly?
The thing that I really appreciated of all five of them is the fact that they are trying to apply AI into the industry. They are choosing, we could always debate whether those are the right problems to solve or not, but the fact that they are very deliberate about trying to solve industry-specific problems just gives me the confidence that there are legs to what it is that they're trying to achieve.
And for me, the thing that stood out is that each one of those solutions aims to empower a human. It's all AI augmenting a human. It has not been about AI replacing a human, but it's about how do I make people better at doing their jobs? How do I make life easier for a customer? So that really stood out, and these were really great examples of what we've been talking about, but building that human plus AI workforce of the future.
And I think what's interesting about what I've been seeing among start-ups and scale-ups in Silicon Valley, only a year or so ago, there was a lot less emphasis on trying to make stuff work in enterprises. And now folk are realising that it's a really, really important part of making money. And I'm hearing Services-as-Software fed back to us from stages in the Valley. We've seen Sequoia recently written a paper admitting finally that Services-as-Software is the future. And we're kind of, yeah, yeah, we all know that, right? We've been saying this for the last couple of years. So it's good to see California catching up in the end, isn't it? So, folks, I'm just going to give one last call for votes. Is anyone stuck trying to vote right now that I should wait a moment? Okay, so we're going to call that. Before we get too excited about that, I do want to do the presentation of the awards and can we get pictures of the guys. I think let's just do that in alphabetical order so I'm going to call you up from the left of the stage now. First of all, big hand for Covasant. Congratulations, come on in, sharks. Thank you. And next on our list is KnowledgeLake. Congratulations. And Binny and Kognitos, I don't know whether you both want to come up but we have just the one trophy. Well done. Thank you. And then we have Siva from Lyzr. Well done, sir. Well done. And finally, our final award is for OrbitShift. Congratulations. Thank you. Thanks, guys. Okay, so just in summary, congratulations to all of those guys. I hope you agree. They did a great job of entertaining you for a little while today. I hope you've all felt a bit informed about something that's coming next, maybe. They're all here, they're all going to be hanging around for the rest of the day. Many of them have booths here as well so please do go and check them out. Find out a little bit more about them and if you have further questions that our sharks didn't cover please go and talk to them about that. And I think you saw on the data, I think the most investable at this time I think was Lyzr so another little congratulations to Lyzr right now. Well done, folks.
If you don't have an account, Register here |
With the exception of our Horizons reports, most of our research is available for free on our website. Sign up for a free account and start realizing the power of insights now.
Our premium subscription gives enterprise clients access to our complete library of proprietary research, direct access to our industry analysts, and other benefits.
Contact us at [email protected] for more information on premium access.
If you are looking for help getting in touch with someone from HFS, please click the chat button to the bottom right of your screen to start a conversation with a member of our team.
