Recently KOGO, a deep tech AI platform, held an event in Goa where almost 27 Indian startup founders were trained to build on KOGO’s small action model platform for AI. The company also announced its plan to build free AI agents to increase the AI adoption rate among SMEs and enterprises.
Both Praveer (Praveer Kochhar, CPO & Co-Founder of KOGO) and Raj (Raj K Gopalakrishnan, CEO & Co-Founder, KOGO) have been serial entrepreneurs. Praveer had a lot of experience in retail technology as the first company he started in 2005.
Along with the technology, Praveer always had a passion for travel and motorcycling. So I invested in and ran xBhp, which is India’s largest motorcycling community. He ran that magazine for six years as a publisher.
So how did they come up with KOGO?
“We were both part of a motorcycling group, used to ride together. And we thought at that time that what if there was something like an AI that could track our route and write a travelogue for us,” he shares.
It’s a virtual assistant that travels with you, takes notes of wherever you stop, whatever you do, what photos you click and turns it into a travelog. “This is a little before the whole LLM, GPT story came about. We got to work and launched KOGO in 2020 as an AI travel storyteller for people who are into road trips.”
In a conversation with CIO&Leader, Praveer Kochhar talks about the past and the future of KOGO. He also shares his vision for the company and the challenges he thinks the tech industry will face when it comes to cybersecurity.
CIO&Leader: What does KOGO mean to you? How does it come to be?
Praveer Kochhar: The idea was that it’s like Strava but with AI where you can track your trips and tell a story. February 2020 is when we launched and March COVID hit and we had to shut shop for a year. We had raised some 300-350 thousand dollars before that to develop the product and then as we came out of COVID we had to look at a different strategy.
We started licensing our stack to auto companies and realised that there was a huge need in the auto companies, especially motorcycle and car OEM market, to build communities around their brands. And we had a great tool that could build communities for travelers. So as they travel and you buy a motorcycle or a car, you have a companion app that they give you. Currently, the companion app didn’t have any additional community features.
Our product immediately fit in very, very well there. We signed up with Bajaj. Six months later, we signed up with CLPL. In ‘22 we worked with MapMyIndia on a project, which was with Mahindra itself. And they took note of the fact that we had a product that OEMs wanted. They were already working with about 30 plus OEMs because they have embedded maps. So it was a good strategic partnership. And we signed a deal and they invested in KOGO.
They’ve already done two rounds of investment and we used that money and we said, okay, we’ve got something in AI and travel. Let’s complete the picture in AI and travel and add the entire suit of AI travel solutions for our OEM partners and as well as MapMyIndia’s D2C brand, which is Mappls. So we added hotels, we added AI travel companion, we added flights, local experiences, built it into an entire, everything you need in travel driven by AI.
Why we were building all of this? We had a good opportunity of, we built a lot of tools that we use to build these AI solutions, and we thought, why not use these tools and put them together into a platform? And this is the same time last year when GPT also came out. We kind of figured that there is going to be a large amount of apps that will be powered by AI.
But the tools to build those apps didn’t exist. So we started working on KOGO OS in parallel and KOGO OS is now what we are taking to many other sectors other than travel. Travel we have a great foothold between us and MapMyIndia. We are now taking steps into other sectors and helping them build apps and solutions foundationally with AI.
So that’s kind of the journey of KOGO. Personally, mixes travel and technology which are two of my biggest passions. One fuels the other and that’s what it means to run KOGO. Never a dull day.
CIO&Leader: What are the challenges you faced while building KOGO OS? And what makes it special?
Praveer Kochhar: OpenAI, Anthropic, Google, by developing these large language models, they’ve basically given intelligence in an engine. So now you have intelligence available as an engine. With an engine, you can build a bicycle, a motorized bicycle, a motorcycle, a car, a spaceship, a train, you can build anything. But the engine is at its core.
So, what what was needed was very clear from day one, the moment GPT 3.5, GPT 4 came out and then subsequent many other language models that came out is that these engines need to be put inside a body and converted into a vehicle that can take you from point A to point B. And that vehicle is what we call agents, AI agents.
Agents are essentially anything that needs to be automated in a workflow. The language model itself cannot do it. It needs tools to interact with the physical world and to automate a certain workflow.
So we had to work extensively on A, building these tools from scratch and making sure that these tools are flexible that they can plug into any kind of enterprise, any kind of existing system. That was challenging.
And B, the biggest challenge that we continuously work on day in day out is how to reduce hallucinations and make sure that these language models stick to the path. Because any enterprise that you work with, the biggest requirement that they have is, oh, the output is not consistent, we want consistent output, we want secure output, we don’t want hallucinations, we don’t want all these things.
So that is the second big challenge that comes working with these language models. They’re also in a very early stage today. They will improve dramatically and they are improving dramatically, but these two challenges are critical.
CIO&Leader: And what about cybersecurity? Is that also one of the challenges you are facing?
Praveer Kochhar: I’ll break down cyber security into two aspects.
One is an external threat. And second is internal threat. When I say internal threat, I essentially mean how to secure the data for an enterprise from going out. Now, we are working with BI companies that want to analyze financial statements. Not only theirs but their competitor’s publicly available content. But the analysis is something that they want to keep to themselves.
Now the biggest challenge in today’s technology is that this analysis requires a large language model to do that analysis. Which means the data has to go to the cloud. So one thing that we are working on heavily and also investing heavily in is making sure that we are able to deliver the same quality with a small language model that can be hosted inside the cloud of the enterprise. Which means the data does not leave your premise and that I think is really the future.
We are very strongly banking on the fact that small language models, a cluster of experts rather than one big blue whale that is an expert on everything is the future. And that solves a lot of concerns that enterprises have, which they want to do local hosted services. They don’t want they want it on the edge.
We are also working on very small models that can be deployed on the edge. So that takes care of internal threads. Like internal threads so that it is a data does not leave my Enterprise my four walls when it comes to external threats, external threats are easily handled more or less by making sure that we are compliant with the top three or four cloud solution providers. So your AWS, we are hosted on AWS, we are compliant with Azure and Google Cloud. We make sure that any enterprise wants to deploy our solutions is available on these clouds and they already have their security measures in place.
We allow large enterprise to host KOGO OS within their cloud ecosystem, which means it already qualifies their security measures. It’s sitting inside their walls and anything that they do in terms of resultant data is also stored within their premise. But obviously, it’s costly and only limited for large enterprise. And I think cybersecurity is also a large concern for large enterprise. And that’s how we try to manage it.
CIO&Leader: Is there any specific audience or sector KOGO is targeting?
Praveer Kochhar: We have to have our focus in place. Our journey, the way we started out was building a product, a very specific market with AI at its core, which is what we built in travel and the OEM space. We had inroads, we are revenue positive, we get paid for our platform, we have high stickiness, we’ve got renewals from our partners. They swear by what we have deployed. But as we look at this, this is one product we have developed using AI. But in any industry, new products can be developed using KOGO OS.
Easiest way to define audience would be who are these people who will build using KOGO OS and define that into three categories. The first category is SMEs. SMEs who come to KOGO OS have some ready built large use case agents which they can customize instantly without any knowledge of coding. That’s the largest base.
The second is developers. People who are developing solutions for other industries and have become domain experts because let’s say there is a there is a IT company that develops insurance products. They would use KOGO OS to build AI into their solutions using KOGO OS. So they would automate certain workflows offer insights additional AI features into their existing product streams. They don’t have to go back and build these tools.
The third audience that we look at which are large system integrators. We are already in conversations with two of the three largest system integrators in India who have large Enterprise clients. These system integrators is who we want to drive adoption of KOGO OS so that they build with KOGO OS and they create their own AI applications. We want to make sure that we remain builders of the OS. We make it stronger, give it more tools so that more and more applications can be built faster deployed anywhere and be managed at scale.
CIO&Leader: How do the upcoming years look like for KOGO?
Praveer Kochhar: We already have a very large pipeline when it comes to the OEM business. We have three of the top four motorcycle OEMs as our clients. We are already on the verge of closing three more car OEMs. Our travel and OEM business is going strong and we’ll see double growth in the coming year.
When it comes to AI, our focus is to look at not only India but MEA also as a market. We’ve already signed up a large distributor for the Middle East market. We are heavily looking at onboarding system integrators and developer partners. So the AI business is all about building a community of developers around KOGO OS.
The target is to have at least 25 serious IT developer teams over the next six months to start building with KOGO OS and at least three to four system integrators that we want to onboard this year who will build for Enterprise. And I feel we are right on track to be able to do that.
CIO&Leader: What is your perspective on the future of AI in the Indian market? While numerous studies and use cases exist, the practical applications seem limited. Additionally, how does KOGO, AI, and OS compare to its competitors in this evolving landscape?
Praveer Kochhar: We have competition. It would be unfair to say that we don’t. A competition is at different stages of development. This is a very new market. There are platforms like Lang Chain, Llama Index, even Microsoft has a platform called Autogen, and KRU.AI.
A lot of these are frameworks on which you build real world AI agents and workflows. What they don’t have today is access to industry and real world use cases. We are already implementing a lot of use cases over the next four to eight weeks. We’ll have things in the market that are being production tested right now.14 POCs going on and in next three months, we’ll have about seven or eight of them already running life. Now, I’ll give you examples of what these use cases are.
We are working with a manufacturing company who want to automate basic content and queries across their C-suite of organizations. There are about 48 separate use cases that they have listed down that they want to answer on a day-to-day basis and we’re just making those integrations in place and getting those answers for them.
Another product that we are working on with a very senior marketing consultancy, performance marketing company is for Amazon sellers. This AI understands reviews, interprets those reviews and turns it into a digestible report.
AI can be your eyes and ears that works 24*7. Save you time in finding information. Distill that information into an actionable item so that you can take faster decisions.
And that’s the core of what we are building. Saving people time doing repetitive tasks so that they can take faster decisions. And I personally believe this is this is great because a lot of time is spent just sifting through information.
CIO&Leader: Given the increasing complexity of AI applications, how do you envision AI transforming the workforce? While some enterprises are focusing on re-skilling and ensuring new hires possess a fundamental understanding of AI, there are concerns about layoffs and a lack of interest in re-skilling current employees. What is your outlook on the future of AI in relation to workforce development and job displacement?
Praveer Kochhar: I often say this and it’s a line that I swear by, that people with AI will replace people without AI. I think as industry leaders or people who are working in this domain, our focus should be, like you said, reskilling to make sure that people adopt and understand AI. Right. And and more and more capable and more useful on a day-to-day basis.
That’s the lens at which we should look at this challenge. Like computers came and people needed computer institutes, App Tech, I went to App Tech to learn Microsoft Word and basic programming long time back. That reskilling is going to be again a very huge market. And people will have to go through it, adopt this new tech.
And yeah, I think it’s going to create superpowers for a lot of people, create companies that take faster decision, move faster, react faster to changes and are so much more well-informed.
Adoption and reinstating should be a primary concern, especially for people who are comfortable in their job. I would say, like it’s a daring remark, but people who are comfortable might not, you know, take in AI as a helping tool, but rather as an opponent.
And I think the goal at KOGO and I think we’ve coined that line as we started thinking, how do we market KOGO OS is human potential meet AI. And that’s what it says on our website also. We believe that there is a potential to bring, for the first time in humanity, a work-life balance. And AI can be that catalyst to bring work-life balance in various industries very quickly.
That’s what I feel we are here for. We have to imagine a better future with technology. And that’s the core of building any piece of tech, right? If it’s not building a better future, A, it will not get adopted because people will reject it. And B, it will struggle to scale beyond a certain point.
CIO&Leader: In your opinion, where is India falling short in the AI sector? Despite our rich history as a fertile ground for startups, we’ve only seen one unicorn this year. We need more startups, especially in emerging technologies, yet many seem unable to reach the finish line. What are the critical gaps hindering the growth of AI startups in India, and why isn’t there more support pouring in for these ventures?
Praveer Kochhar: I think AI is in its early days. The more I’m talking with people, businesses, startups, and the more we are going in industry conferences, there is a lot of work that is happening in AI. I think give it another six months, this industry in India will also reach a certain amount of maturity. A lot of these tools and technology is very new.
A lot of money is pouring to the source. And by the source, I mean where these models, these core intelligence models are being born. These are being born in the US with OpenAI, with Anthropic. So that’s where a lot of money is being funneled because if you own the engine, you control what the engine builds tomorrow. So that’s why a lot of money is going there. But I also feel over the next few quarters, a lot of current industry leaders, India is number two in SaaS in the world.
A lot of these homegrown SaaS products are very rapidly adopting AI. And I think that will be the leapfrogging stage for them to go to the next level of investment and attract that scalability because applications with AI are yet to be seen. They are being built as we speak.
And that’s the whole idea of why we are also in this whole game of Google OS because we know everyone wants to build with AI today. But the tools are not there. So give it a few quarters. I think there is a maturity that this market needs to achieve and money will flow in.
CIO&Leader: India has been in the top in adopting AI as a tech but I think we do need more people behind the scenes who are working on it continuously and seeing more unicorns grow up as well. Should startups be concerned with cyber security?
Praveer Kochhar: Cybersecurity in the world of AI means very different things and a lot of us are only beginning to learn what that landscape will look like. We are yet to see how AI is going to manage cyber security. Because there will be a counter of that AI that is going to break into cybersecurity. So I think those tools and that landscape is not yet clear.
As startups using AI, I would say it is firstly very important to understand the potential and the risk of using this technology. And when I say that, I essentially mean that it’s something that can auto-execute code. It cannot execute and create content. It can start building spam.
So a responsibility framework around building with AI, which is something that we are very big on. We wrote down our own manifesto before we started our website because we said there have to be principles of AI.
I think what startups need to understand is draw certain guardrails, live by those guardrails, standard guardrails of AI, while building anything with AI, that we will not cross these guardrails. Ethical responsibility is far bigger than the responsibility of the tools, because eventually that dictates what we implement in the real world.