While Silicon Valley debates the next ChatGPT breakthrough, Manas Bhasin and Aditya Singhal are quietly keeping the AI economy running—processing over 1 million AI operations every single day for 50+ startups who’d collapse without them
Bengaluru, India – August 29, 2025 – Here’s the story that won’t make TechCrunch’s front page, but it absolutely should.
While everyone obsesses over the latest AI model that can write Shakespeare or solve calculus, two 22-year-old founders in India are solving the problem that keeps AI startup CEOs awake at 3 AM: how to make their brilliant algorithms actually work for real customers without going bankrupt.
Meet the invisible power brokers of the AI boom. Manas Bhasin and Aditya Singhal run Metabees, and their platform Echidna AI Foundry has become the secret weapon separating AI companies that ship from those that perpetually demo.
They’re processing over 1 million AI operations daily across 50+ startups—the mundane, crucial, money-making operations that happen every time someone uses AI and expects it to work.
Here’s the kicker: In an industry where 70% of AI projects never make it from demo to deployment, Echidna AI Foundry boasts a 94% success rate. That’s not just impressive—in enterprise software, it’s practically supernatural.
The Real Problem No One Talks About
“Everyone’s building for the AI companies that get TechCrunch covers,” says Manas, leaning back in their Bengaluru office. “We’re building for the other 95%—the ones trying to ship fraud detection that doesn’t crash, diagnostic AI that passes FDA approval, or recommendation engines that scale without eating their budget alive.”
Both founders lived this nightmare in previous ventures. That shared trauma became the foundation for what they’re building now—infrastructure so reliable that their clients achieve 73% faster deployment and 68% lower operational costs, with 89% renewal rates and 60% referral-driven growth.
The Stories That Matter
Take the fintech startup burning 65% of engineering bandwidth just keeping their fraud detection
AI alive. After Echidna AI Foundry: same team, 10x transaction volume, 23% better accuracy.
“We track ‘engineering efficiency’—how much time startups spend on core innovation versus operational overhead,” Manas explains, pulling up client dashboards. “Before us: 35%. After: 78%. When you multiply that advantage over months, these companies iterate twice as fast as competitors.”
The multiplier effect is staggering. Each client serves thousands of customers, meaning Metabees indirectly powers AI experiences for nearly 10 million end-users globally. When they optimize a fraud system, millions get better financial security. When they stabilize diagnostic AI, thousands get better healthcare.
Perfect Timing for an Unsexy Business
While AI funding hit $25 billion in 2024, the party’s getting more selective. Investors now ask hard questions about unit economics and operational scalability. The winners won’t have the fanciest algorithms—they’ll have the most reliable operations.
“There’s this massive shift happening,” notes Aditya, who handles business with systematic precision. “Two years ago, ‘GPT’ in your pitch got you funded. Now investors want proof of sustainable competitive advantages. Operational excellence is becoming the differentiator.”
Building the Boring Stuff (That Actually Matters)
What’s refreshing about their approach: it’s deliberately unsexy. No foundation models, no AGI promises. Just making existing AI systems work better, cheaper, more reliably.
“Everyone wants to work on cutting-edge research. Nobody wants to build the infrastructure that makes breakthroughs useful,” Manas reflects. “We’re happy doing boring work because boring work is where real business value lives.”
Their team doesn’t publish NeurIPS papers—they ship features that measurably improve client lives. The result: a platform that might not win design awards but consistently delivers business value.
“We joke that we’re building AI industry plumbing,” Manas grins. “Plumbing isn’t glamorous, but when it works, everything works better. When it breaks, nothing else matters.”
What’s Next now?
At 50+ clients processing over 1M+ daily operations, Metabees isn’t celebrating—they’re just getting started.
Bhasin and Singhal have their sights set on something far more ambitious: building the universal infrastructure layer that will power every AI deployment on the planet.
While the rest of the industry chases the latest model benchmarks and flashy demos, Metabees is quietly building the unglamorous but absolutely critical foundation—the pipes, the reliability, the scale that makes everything else possible.
This isn’t about incremental growth or riding the current wave. Bhasin and Singhal are positioning Metabees to become infrastructure—the essential layer that sits beneath every breakthrough, every deployment, every company serious about shipping AI that actually works.
The timing couldn’t be better. The world is rapidly moving toward an AI-first everything, and someone needs to own the foundation. Metabees believes that someone should be them
For two 22-year-olds, it’s remarkably mature strategy for a market desperate for practical solutions. The AI revolution isn’t just about better algorithms—it’s about better infrastructure. And increasingly, that infrastructure has a name: Echidna AI Foundry.
The next time you use AI-powered fraud detection, get an AI medical diagnosis, or receive AI-driven recommendations, there’s a decent chance two young founders in Bengaluru are the reason it actually works.
🌐 Website: www.metabees.org
