AI assistants are evolving from single‑skill chatbots into networks of specialised agents that can cooperate, share memory and execute end‑to‑end workflows. Few companies illustrate that leap better than KOGO Tech Labs, the Bengaluru‑based startup behind KOGO OS and its new “agentic mesh.” Debuted at this year’s Auto Expo, the mesh marries on‑prem language models with MapMyIndia’s granular geo‑data to create a universal in‑car assistant—one that not only controls vehicle functions but also books deliveries, reads email and hands conversations from a weather agent to a cooking agent without losing context.
In a recent conversation with CIO&Leader, Raj K. Gopalakrishnan, CEO of KOGO Tech Labs, explains why data‑sovereign deployments matter, how fine‑tuned micro‑models keep user information private, and where the mesh will land next—from banking SOPs to healthcare TPAs. He also outlines revenue targets of $5 million this year and $60 million within three, candidly comparing the adoption climate in India and North America and advising CIOs on how to move from AI flirtation to full‑scale transformation.
CEO, KOGO Tech Labs
CIO&Leader: In light of prevailing data‑privacy concerns, what safeguards has KOGO Tech Labs put in place to protect user information?
Raj K. Gopalakrishnan: Our strength is running on‑prem. The core layer resides in the OEM’s data centre, and the operational layer runs on the edge inside the vehicle, so nothing sensitive goes to the public internet. Publicly accessible apps still connect outward, but all personally identifiable information stays local.
We never send raw data to a large public LLM; that would expose the user completely. Instead, each task relies on a fine‑tuned small language model that lives on‑prem. One model handles weather, another cooking, another travel, and so on. Because each model is narrowly scoped, privacy is preserved.
CIO&Leader: Could you elaborate on the strategic rationale behind your partnership with MapMyIndia?
Raj K. Gopalakrishnan: MapMyIndia is India’s leader in geo‑intelligence. For a mobility‑focused assistant, granular map data is essential, and they have village‑level roads, ISRO feeds—depth no other provider, not even Google Maps, can match here. Integrating their datasets with our agentic mesh was the obvious choice for an automotive assistant.
CIO&Leader: What tangible benefits do you anticipate the universal in‑car assistant will deliver to end users?
Raj K. Gopalakrishnan: We want the car to become another person in your life—navigator, travel buddy, concierge and personal assistant rolled into one. Through natural language, drivers should access any capability they expect from a modern vehicle and its connected ecosystem.
CIO&Leader: When can consumers expect to encounter this assistant in commercially available vehicles?
Raj K. Gopalakrishnan: We announced it two days ago, and we’ve been testing with several OEMs—names under NDA—for six months. Adoption depends on each OEM’s launch cycle, but many new models coming out toward the end of this year should feature the technology.
CIO&Leader: What revenue objectives has KOGO Tech Labs set for the current fiscal year and the subsequent two to three years?
Raj K. Gopalakrishnan: Because we deploy on‑prem, we operate on a licence‑plus‑consumption model. Our focus for this fiscal year is about US $5 million; over the next two to three years we expect to surpass $60 million.
CIO&Leader: Beyond the automotive sector, in which additional industries are you deploying the agentic mesh?
Raj K. Gopalakrishnan: Plenty. The agentic mesh applies to any industry. Our largest deployments are in banking, insurance, healthcare and defence. Take banking: a simple request like “increase my credit‑card limit” triggers a complex SOP—credit‑worthiness checks, CIBIL score, payment history, income verification. Each step involves a decision point. We replace those manual steps with agents that talk to each other and arrive at the same decision autonomously. A typical bank has 135–140 such SOPs; all can be agent‑driven. Similar opportunities exist in insurance claims, healthcare TPA workflows and more. Most of these implementations are in North America.
CIO&Leader: What specific challenges do you face when deploying your solutions in India compared with North America?
Raj K. Gopalakrishnan: U.S. enterprises have well‑defined processes and a higher cost of labour, so ROI is clear and adoption is rapid—about 70 % of global implementations happen there. Indian enterprises want AI, but their SOPs are less mature and manual labour is cheaper, so decisions take longer. Without a well‑defined process, AI transformation is hard; you can’t automate what you haven’t mapped.
CIO&Leader: Would you outline the company’s strategic roadmap for 2024–25?
Raj K. Gopalakrishnan: We’re concentrating on on‑prem deployments in BFSI, healthcare and defence, where data privacy is critical and processes are well documented. Simultaneously, our innovation lab in Cochin keeps iterating on the agentic mesh; it took 10–12 months of R&D to reach this point, and we’ll keep pushing the technology forward.
CIO&Leader: What guidance would you offer CIOs who are evaluating the adoption of AI agents within their organisations?
Raj K. Gopalakrishnan: First, make up your mind. Commit to credible AI transformation. Identify the process you want to improve and gather the relevant data. Know why you’re doing it—efficiency, cost savings, customer satisfaction—and define success metrics. Then approach an AI partner. Asking, “What can AI do for me?” without that groundwork only wastes time.
CIO&Leader: What factors influenced KOGO Tech Labs’ decision to introduce its technology at Bharat Auto Expo?
Raj K. Gopalakrishnan: You already know what we do—KOGO is an AI‑agent platform. At its core is the KOGO OS, an operational layer on which our AI agents run. Traditionally, agents possess decision‑making and reasoning capabilities but can handle only single, isolated tasks. We’ve been working on—and have now launched—an agentic mesh built on a universal agentic protocol. This lets one agent hand a conversation, complete with context and memory, to another agent.
Auto Expo is the first public showcase of that mesh. By fusing it with MapMyIndia’s deep geo‑intelligence, we’ve created a universal in‑car assistant. Unlike typical voice assistants that merely open apps, ours can operate them. Ask it to navigate from Chirag Delhi to Malviya Nagar and it starts the route, not just the navigation app. While you’re driving you can ask, “What’s the weather in Delhi tomorrow?”—and get the answer—then immediately say, “Teach me how to make sushi.” The weather agent realises it lacks that knowledge and seamlessly passes the conversation to the cooking agent, transferring the entire context.
Next you might ask, “Will Virat Kohli stay on the Indian cricket team?”; the system hands off to a sports‑analysis agent, which can then shift to a personal‑assistant agent to read your email, or to a travel agent to plan a three‑day trip. This criss‑crossing of tasks is breakthrough technology. From checking your bank balance to ordering two kilos of potatoes on Instamart, every app and service becomes voice‑operable inside the car.