Evozn.ai started in March 2025 as a side project between Founder Ram Raj Bista and Co-Founder Naresh Gowda — two people who couldn't stop thinking about one broken demo. Eighteen months later, it's ten engines running live inside a real product, and we're still the only two names on the cap table.
In March 2025, Founder Ram Raj Bista was building the first version of RiseYou — an idea simple enough to fit on a napkin: let a voice AI conduct real job interviews, so hiring teams could talk to more candidates without burning more hours. The prototype worked, right up until it didn't. Left alone for more than a few minutes, the model repeated a question it had already asked, skipped one it hadn't, or raced through a closing script before a candidate got the chance to answer. It wasn't a small bug. It was the whole idea, quietly failing in the one place it couldn't afford to.
There was no seed round to fall back on and no team to hand the problem to — just a laptop, a deadline that existed only because he'd set it himself, and Co-Founder Naresh Gowda, who looked at the same failure and saw an engineering problem instead of a dead end. What followed wasn't a pitch deck. It was months of late nights, the two of them rebuilding the same interview loop over and over, testing it against people it was never designed for — an aerospace engineer, a farm supervisor, an ICU nurse — because if it couldn't survive someone outside the demo script, it wasn't ready for anyone.
Nobody funded that work. There was no runway to protect, so there was nothing to lose by being honest about what wasn't working yet — which, it turns out, is the only way to actually fix something this stubborn. Piece by piece, the fixes Ram and Naresh made stopped being patches and started being a real layer underneath the model: something that held a capable AI to an exact standard without making it sound scripted. That layer became Evozn.ai. It shipped inside RiseYou first, because that was the only place it could be tested against something real, and it's been in production, growing, ever since.
We're writing this in August 2026, eighteen months after that first broken demo. It's still Founder Ram Raj Bista and Co-Founder Naresh Gowda, no outside investors — every engine on this site was paid for with the same stubbornness that built the first one. That part of the story hasn't changed. We just hope it's the least interesting part of what we build next.
RiseYou's first voice-interview prototype fails the way generative models fail — plausibly, not obviously. Ram Raj Bista and Naresh Gowda decide to fix the model's reliability themselves rather than ship around it.
No outside funding, no hires — just the two of them rebuilding the interview loop until it held up against real, unscripted conversation instead of a rehearsed demo path.
The reliability layer goes in front of candidates it was never designed around — aerospace, agriculture, healthcare — because a system that only survives the easy case isn't done yet.
What started as one reliability fix is now ten engines running live inside RiseYou, still self-funded, still built by the same two people who started it.
Every claim on this site is something we tested, not something we project. If it's still direction rather than a shipped result, we say so — see the roadmap note under Architecture.
RiseYou isn't a customer we designed around after the fact. Every engine here was shaped by a real, specific need inside RiseYou's product first — the use case came before the API.
A model that never repeats a question is a less exciting demo than one that improvises brilliantly. It's also the only one you can actually put in front of a real candidate.
Voice interview AI is usually shown off on software engineers. We deliberately ran it on industries it wasn't built around first, because that's where a resume-anchored system actually breaks.
RiseYou is where every engine on this site is proven first, against real interviews with real candidates — not a sandboxed demo.