In an interview with TimesTech, Rajesh Gupta, CEO & Co-Founder, Metaculars traced his shift from big tech to entrepreneurship, the pivots that shaped Metaculars, and the lessons that guided its eventual acquisition by Skan AI. Rajesh shared how staying close to real user pain, reading market timing, and unlearning perfection helped the company find its direction. He also explained how Agentic AI is set to reshape how businesses act, decide, and operate.
Read the full interview here:
TimesTech: You’ve worked with some of the world’s most advanced AI teams at Apple and Qualcomm. What motivated you to leave that environment and build Metaculars, and how did that transition shape your approach to innovation?
Rajesh: I loved my time at Apple and Qualcomm. Great teams, meaningful work. But over time, I felt too far from the real user. I wanted to build something end-to-end, where I could see a user struggle, ship a change, and watch it genuinely help them. That push to be closer to the outcome is what nudged me toward entrepreneurship.
The transition was not sudden. I spent almost two years preparing, getting my personal life in order, picking up the right experience, and finding the right cofounder. That changed how I think about building- Less theory, more truth. Start with a real customer pain point, not the technology.
There was a lot of unlearning too. Big companies train you to perfect things. Startups make you move fast. In large organisations you worry about your own part. In a startup you worry about everything, from money to hiring to direction. Mentors and advisors made a big difference there.
Metaculars came from passion, the desire to build something meaningful and grounded in real problems. That mindset still guides how I build today.
TimesTech: What was the original insight behind Metaculars, and how did it evolve into a product compelling enough for Skan AI to acquire? What were the key turning points in that journey?
Rajesh: Metaculars actually started with a very different problem. My cofounder Sumit and I were looking at the day to day pain data science teams faced, messy ML workflows, scattered assets, and no proper system of record. Then ChatGPT arrived and changed how data teams approached their work it became clear our original idea was built for a world that was already shifting, so we had to rethink.
We explored LLM evals next. It looked promising at first, but the more conversations we had, the clearer it became that the space was getting crowded fast. MLOps companies were moving into LLMOps, investors were not excited, and customers saw evals as something to deal with later. It was not a bad idea, just the wrong timing.
That pushed us to move up the stack. Instead of building tools, we focused on a real business problem. Companies were building AI features and running PoCs, but user adoption lagged. Especially in SaaS. Onboarding was slow, and products did not guide users well.
That insights shaped our final pivot. We started building GUI agents that help users take the right actions instead of hoping they figure it out.
This direction aligned with where Skan was already heading. They had a strong base in process intelligence and were exploring agentic AI. Our work with GUI agents fit naturally into that vision and eventually brought both paths together.
TimesTech: Every startup journey teaches something hard-earned. What were the biggest lessons you carried forward from building and exiting Metaculars — especially when it comes to product-market fit and timing?
Rajesh: Metaculars taught me that PMF is mostly about honesty. You can love your idea or the tech, but none of it matters if the customer’s pain is not real enough. Once we shifted to less theory and more truth, things became clearer.
Timing was another big lesson. When ChatGPT launched, the market changed almost overnight. What made sense in October felt outdated by January. It taught me to treat timing as something you read continuously, not something you assume stays the same.
We also learned the difference between an important problem and a paid problem. LLM evals sounded meaningful, but customers treated them as an afterthought and investors were not interested. That pushed us to solve something people actually felt, which was slow onboarding in SaaS.
Speed was another survival skill. Coming from big companies, perfection becomes a reflex. In startups, it slows you down. You have to unlearn a lot to match the pace the market demands.
And PMF is not a finish line. Enterprise AI is still early. It is too soon for anyone to claim they have nailed it. We saw early signals, but it was still developing.
TimesTech: You’re now leading Agentic AI at Skan. How do you define Agentic AI, and in your view, how will it change the way businesses operate and make decisions over the next few years?
Rajesh: I explain Agentic AI in a simple human way. It looks at the world like our eyes, learns like our brain, acts like our hands, and needs trust the same way we trust a teammate with the right guardrails.
What makes it different is that it does not stop at showing insights. It closes the loop. Instead of saying Here is a problem, it fixes it, adapts when things change, and tells you what it did.
For business teams
People spend hours staring at dashboards and still debate what to do next. With agentic systems, that back and forth shrinks. The AI will surface the signal and also take the next best step, like rebalancing a campaign, sorting leads, or catching churn risks early.
For IT and engineering
A lot of time goes into glue work like integrations, handoffs, and small workflow triggers. Agents remove that load so teams can focus on bigger decisions.
For operations
Instead of stitching together twenty rigid steps, you will have agents that understand intent and adjust as things change. If there is a supply chain delay, the agent recalculates inventory, updates procurement, informs customer success, and adjusts timelines without people rushing into a Slack war room.
The shift is simple. Businesses move from insight-driven to action-driven. And once you see that loop working, it is hard to go back.
TimesTech: What are some practical insights you’d share with founders trying to build and scale AI-driven startups today — especially given how fast the ecosystem is evolving?
Rajesh: Start with the customer pain, not the AI.
Many founders begin with the idea of building something in AI and then try to find a problem. That rarely works. AI is just a tool. If you do not anchor yourself in a real pain point, the company ends up chasing hype instead of value.
Do not fear pivots, but avoid knee jerk ones.
There is a thin line between moving too fast and staying stuck. Founders get attached to their first idea because they live with it every day. Stepping away for even a day helps you see the truth more clearly.
Ask for help.
Founding is lonely, and most things happen for the first time. Lean on people around you. Nobody builds a company alone.
If possible, start with a cofounder.
Being a solo founder is extremely challenging. You can build the product, but a startup is bigger than the product. Sharing the emotional, operational, and strategic load makes a huge difference.
Remove auto login from Twitter and LinkedIn.
It sounds small but it is not. Auto login pulls you into social media every few minutes. Turning it off protects your attention, which is one of your most valuable assets.
TimesTech: There’s a lot of hype around AI. How do you separate the noise from real value creation? What principles guide you when designing AI systems that genuinely move business metrics?
Rajesh: For me it starts with one question. Is someone willing to spend their time or money on what we are building
Real demand does not come from hype. It comes from a customer who adopts, integrates, or pays. That is the cleanest signal.
As a small startup, you also can’t (and shouldn’t) try to build the full product in month one — or even year one. The trick is figuring out which slice truly matters. Ignore the buzzwords, ignore the pressure from incumbents, ignore every big launch from OpenAI for a moment. Focus on the piece of value that a customer notices immediately.
If your market is large enough, your reasoning is grounded in first principles, and customers are giving you their time or their wallet, you are on the right path. Everything else is noise.
