KOGO’s AI Shift: From Travel Stories to Enterprise Intelligence

Praveer Kochhar, Co-founder & Chief Product Officer at KOGO AI, bringing over 20 years of experience across diverse sectors including retail and technology. Leveraging his background in community building and applying technology to the travel space, Praveer co-founded KOGO Tech Labs with Raj K Gopalakrishnan. Together, they created KOGO – India’s only AI-powered Travel Expert App.

KOGO has since evolved beyond its original travel-focused mission to become a platform that empowers developers and enterprises with universal AI agents.

In this exclusive conversation with CIOL, Kochhar reflects on KOGO’s transformative journey—from building a travel storytelling app to developing a secure, full-stack agentic AI platform powering enterprises across industries. He shares how KOGO evolved into a human-first AI company solving real-world friction with Private AI infrastructure, an agent store, and a low/no-code OS that meets enterprise demands for autonomy, security, and scale. Kochhar also offers sharp insights into what defines a truly intelligent agent, why India holds a global edge in agentic AI, and how KOGO is enabling a future where AI doesn’t just assist but acts—with humans in the loop, by design. Excerpts.

If you look at Kogo AI’s evolution - from storyteller to system thinker- for instance, you started KOGO with the idea of intelligent storytelling through travel. Now, you're powering enterprise AI agents across industries. What mindset shift did that journey demand, say, from building an app to building an AI operating system?

When we started KOGO, we were solving a very specific problem—how to automatically tell the story of a travel journey. But as we built that, we realized the real opportunity was much bigger. It wasn’t just about travel but rather about using AI to augment human capability.

As we dug deeper into what was technically possible, we saw the potential for automation in enterprise settings. The possibility of building systems that could reason with data, make decisions, and drive action. Along the way, we started talking to companies of all sizes, and it became clear that there’s excitement around AI, but also a lot of fear. Especially in sectors like finance, healthcare, and defense, where data is everything.

That’s what drove the shift in mindset. We stopped just building smart agents and started building the infrastructure to run them securely. We built KOGO OS, a low-code no-code platform to create and manage agents, and the KOGO Agent Store to deploy AI agents instantly. And we built Private AI, which lets enterprises run everything, be it models, tools, or data, within their own environment.  So the shift wasn’t just in what we were building—it was in why. We’re here to make AI usable, secure, and human-first. That’s what drives everything we do.

If one looks at redefining “agent” in a world flooded with AI tools, with so many companies adding ‘AI copilots’ and chatbots to their stack, how do you define a truly intelligent agent? What must it be able to do before it earns the title ‘agent’?

The term “AI” often brings chatbots to mind, but agents are functionally different. A true agent has cognitive capabilities—it can reason, make decisions, and act independently with minimal human input. What makes agents even more powerful is their ability to learn. As they interact with systems and data, they get better at recognizing patterns, making decisions, and adapting to new situations, without needing to be reprogrammed.

This kind of autonomy allows agents to go well beyond simple task execution. They can manage complex, multi-step workflows, operate across multiple applications, and take on specific business functions. That’s the key distinction. Agents are transforming industries by performing tasks themselves, not just assisting with work.

At ground zero, the real-world is full of friction, polished prompts, often having a placebo effect than a real outcome. Many generative AI tools sound great in demos but fall short in actual business contexts. What are some of the most unexpected real-world hurdles you've had to solve while building deployable agents for enterprises?

One of the biggest challenges we saw early on was that most companies were stuck in POC mode. There was a clear gap between understanding AI conceptually and implementing it in production. Companies were experimenting with prompts, dealing with hallucinations, tweaking guardrails, and struggling to deploy agents at scale in real business environments.

What enterprises really want is AI that works with their existing systems, follows their data policies, and delivers value from day one. They want tools that plug in, automate, and scale. We understood this and built a low-code, no-code platform for creating and managing agents. It includes 600+ pre-built integrations, supports all major LLMs and SLMs, and runs on our in-house agent graph. It’s secure, flexible, and fast to deploy on-prem, in the cloud, or in a hybrid.

With Private AI, our customers own everything—not just their data, but also the models, training, agents, and the resulting IP. That level of control is what makes AI something they can run, scale, and trust. Another major hurdle was cost predictability. Most businesses don’t know how to budget for AI. So we built a token-based pricing model to give them full visibility and control, even when using multiple tools and APIs.

You also underscore ‘the human behind the interface’ or 'the human-in-the-loop'. As CPO, how do you balance building powerful agent functionality with creating trust, transparency, and delight for the end user? What does ‘good product design’ mean in an AI-first world?

Even though we’re building autonomous agents, at the end of the day, these systems are meant to be used by people. So we always start with the core problem—what are we solving for, and how do we make it easier for the end user? That’s what guides our product thinking.

For us, good product design in an AI-first world means realizing human potential with the help of AI while keeping trust, control, and usability front and center. Take Private AI, for example. We built it to remove the biggest blocker to AI adoption: data security. In sectors like defense, healthcare, and BFSI, innovation often stalls because teams can’t risk sending sensitive data to external systems. So we built something they can trust, an AI stack that runs entirely on their own infrastructure. Their data stays with them. Their models, their training, their outputs—nothing leaves.

We’re the only platform in the country offering 100% privately run agentic infrastructure, in partnership with companies like Tech Mahindra and Qualcomm. It’s built to meet strict security guidelines, so enterprises don’t have to compromise on security to move forward with AI.

When we listen closely to what customers actually need, we’re able to build solutions that are usable, reliable, and trusted and that’s what good product design looks like for us.

What is your reading of India's tech edge in the agentic era? Do you think India, because of its diverse data, frugal innovation mindset, and engineering depth, has a unique edge in leading the global agentic AI movement? Or are we still too focused on ‘build to demo’ rather than ‘build to scale’?

India has a real shot at leading in the AI space. The AI talent concentration here has grown massively, over 260%, since 2016, which gives us a strong foundation.  The country is home to 16% of the world’s AI talent, showcasing its growing influence in AI innovation and adoption. We have the engineering depth, the diversity of data, and a growth mindset geared toward unique innovations. That combination can be a serious advantage in the agentic AI era.

But we need to invest more in R&D. In 2024, U.S. private AI investment hit $109 billion, which is almost 12x China and 24x the UK. The world is moving fast, and we can’t afford to lag.

But the momentum is already here, as over 80% of Indian businesses are exploring agentic AI use cases, and the government is actively pushing innovation. Initiatives like AI Centres of Excellence and plans for skill development hubs are steps in the right direction. So, the edge is there—but we need to move from “build to demo” to “build to scale.” Now’s the time to create solutions that are innovative, secure, enterprise-ready, and globally competitive. 

Where do you go from here, can you talk about Kogo’s growth trajectory, and how do you plan to differentiate in the multipolar AI industry and your sweet spot?

We’ve come a long way from solving one specific problem to building a full-stack agentic AI platform designed for real-world enterprise use. And we’re still evolving. We’re actively listening, adapting, and improving the platform to meet what enterprises actually need on the ground.

In sectors like BFSI, defense, and healthcare, public cloud isn’t an option. Most AI tools today rely on open-source libraries and external APIs, which puts data at risk. That’s why we built Private AI that runs on the customer’s infrastructure. The fear around data security is taken off the table. For example, a defense organization can run KOGO entirely on-prem, using local models, with zero cloud dependency. Their agents will function offline, fully compliant with their internal security protocols.

Additionally, most organizations don’t have the teams to build complex AI systems. That’s why our solutions are low/no-code. It can be deployed fast with full control over models and data. Our Agent Store lets teams pick ready-made agents, like a BI agent that lets you talk to your own company’s data and use it instantly.

And then there’s the challenge of one-size-fits-all agentic tools. Most platforms stop at templates, but enterprises have specific, layered needs. For example, if an automobile platform needs to understand in real-time, it can build an agentic system using KOGO OS that pulls data from sensors, machines, logs, and worker reports, all in one place, to diagnose the problem in real-time.

So, we have a clear path forward. Everything we’ve built so far has been shaped by real, pressing problems from our customers, and that’s not going to change. We’re focused on building more autonomous, secure, and scalable agents, making AI usable across more functions. Along those lines, we’re putting a lot of energy into using the KOGO OS engine to build more complex, more autonomous AI systems. Think of a completely automated factory, where all decisions are taken by AI, with a human in the loop to validate. Our core focus is to make AI easy to adopt, easy to build on, and easy to trust. That’s our sweet spot and it always will be.