Sridhar Muppidi of [x]cube LABS on Why He Cut 400 Jobs and Revenue Never Moved
The serial founder behind PurpleTalk and Ello.ai on AI's brutal math, the SaaS reckoning ahead, and the billion-dollar exits he talked himself out of.
[x]cube LABS, the digital agency Sridhar Muppidi co-founded, went from over 1,000 employees to roughly 600. Group revenue stayed flat at $23 to $25 million. A 40% cut in headcount, no drop in output.
Most people would call that a layoffs story. Sridhar calls it a story about agency.
Anybody with high agency is going to survive in this market. Anybody who waits for somebody to tell them what to do is going to have a hard time.
Check out the video of the conversation here or read on for insights.
The people he could not keep, he is careful to say, were not the juniors. They were whoever could not relearn fast enough.
The core competency you brought in has become a commodity. The folks we couldn’t retain are folks who couldn’t change quickly. Not that they won’t be able to change. It’s just they couldn’t change quickly with us.
Sridhar has watched this kind of thing before. He started his first company in 2000, founded Hyderabad’s first Java user group in 1997, and built telecom systems at PanTerra Networks before he ever touched a mobile app. That history is why his read on the present is worth slowing down for. He knows what subsidized booms do.
Right now tokens are subsidized, heavily. They’re like drug dealers, getting us used to it, giving it away for free. But there’s going to be a reckoning soon. We are going to be helpless without those tools.
The new math of building software
The economics today are lopsided. An entry-level engineer in India costs roughly one lakh a month. Enterprise-wide token spend, by Sridhar’s estimate, runs $10,000 to $20,000 a month. Across building and running products, [x]cube LABS already spends more than $1 million a year on tokens. The arbitrage is real, though he insists nobody prices it that way yet, because the gap is still too wide to bother.
What actually changed inside the agency was not “developers use AI.” It was a shift from writing code to orchestrating the agents that write it.
You set the vision, the framework, the non-negotiables. These decisions are mine alone. Those are the things you need to articulate well to any coding agent.
And like any collaborator, the agents forget.
You can’t say, I told you two hours back. The context gets so big they start compressing things. So every one hour or so, you come back to it.
Done well, he says, one person becomes a team of five, or a hundred over two years. Which is also why he expects a reckoning for software vendors. When the cost to build approaches zero, the per-seat subscription gets interrogated.
Somebody charging you $300 a month per seat will get asked: what’s your cloud cost for my 100 people? I’m paying you $30,000. Your actual cost is $3,000? So why am I paying 10x more?
India hosts more than 110 unicorns, roughly half of them SaaS companies. Sridhar expects heavy price pressure across that group within a year. The only durable moat, he argues, is proprietary data, not workflow logic that AI now copies instantly. The bigger prize is a rebuild: for 40 years, humans bent themselves to software’s menus. He thinks that ends, with applications becoming conversation-first and the primary user shifting from a human to an agent.
Right every time, rich almost never
What makes Sridhar a useful narrator is that he keeps being early and selling cheap. PurpleTalk was the first Indian company to ship a game on the Apple App Store. In 2008 it built AdShare, an early iOS ad network with 400 to 500 developers, and sold it for a few hundred thousand dollars, because from Hyderabad, 500 felt small. A near-identical competitor, Pinch Media, had about 40 customers, sat beside Silicon Valley, merged with Flurry, and was eventually acquired by Yahoo for a reported $200 to $300 million.
Then there is Nukkad Shops. In 2015, its CEO Vivek Shukla pitched dark stores and 15-minute delivery, years before Zepto and Blinkit. Sridhar killed it, partly to avoid the capex, partly out of a wish to protect kirana stores rather than compete with them.
The takeaway is chase money. You won’t regret chasing money. Don’t confuse entrepreneurship with social entrepreneurship.
The bet he’s staying in
Now he is building Ello.ai, a voice platform engineered for 12 Indian languages and over 100 dialects, targeting 100 million monthly minutes of enterprise calls by 2026. A sibling product, Upshot.ai, was named Best Customer Engagement Platform, India at the 2024 Martech Awards. His philosophy on voice is deliberately plain: not to sound human, just to work, to read tone, to hold context, to deliver value without pretending to be a person.
For the first time, the founder who wandered off every wave is staying with one.
We’ve tried about 20 things. Six, seven of them are doing quite well. So no regrets that way. But we could have been billionaires instead of millionaires.
Listen now!
Other ways to listen:
Your Feedback matters
As always, I’d love to hear your thoughts! Whether it’s about this episode or ideas you’ve been playing around with, shoot me an email at ad@thepodium.in. Your feedback keeps these conversations going, and I’m always up for chatting about your startup ideas too.
Until next time,
Your Host,
Akshay Datt