The Origin: Speaking Is Natural, Typing Is Not
Humans have been speaking for 100,000 years. We've been typing for about 150. Yet in 2026, the primary way we interact with computers is still through a QWERTY keyboard — a layout designed in 1873 to prevent typewriter jams.
This disconnect is what led Raman Goyal and Himanshu Kumar to build Zavi. The idea was simple: what if your voice could do everything your keyboard does — but faster, across every app, and with AI that understands what you actually mean?
The Founders
Raman Goyal (CEO) studied at the University of Edinburgh and went through both Antler and Entrepreneur First — two of Europe's top founder programs. His background in product and go-to-market strategy drives Zavi's positioning as the Voice Agent OS.
Himanshu Kumar (CTO) graduated from IIT Kharagpur and spent years at Nvidia and AMD working on systems-level engineering. His deep expertise in low-latency computing, signal processing, and cross-platform development is what makes Zavi work across iOS, Android, macOS, Windows, and Linux.
Building Across 5 Platforms
Most startups launch on one platform and expand later. Zavi launched on all five — iOS, Android, macOS, Windows, and Linux — from day one. Why?
Because voice is a system-level input. You don't check email on only your phone or only your laptop. You use both. A voice layer that only works on one platform creates friction — you'd have to switch between voice and keyboard depending on which device you're using.
Building across five platforms as a two-person team was the hardest technical challenge. Himanshu architected a shared core that handles voice processing, AI cleanup, and agent execution, with platform-specific input layers for each OS.
Zero Marketing, 171 Upvotes
Zavi came out of stealth on February 15, 2026. Within 12 days, it earned #7 Product of the Day on Product Hunt with 171 upvotes and 423 followers — entirely organically with zero marketing spend.
The product is rated 5/5 on both the iOS App Store and Google Play. Real enterprise inbound started within a week, with CEOs requesting multi-channel inbox agents and digital executive assistants.
Beyond Voice Typing: The Voice Agent OS
Zavi started as a voice typing keyboard — but that was always step one. The vision from day one was to build the Voice Agent OS: a system-level voice layer that doesn't just type what you say, but understands what you want to do and executes it across every app.
Today, Zavi's four-layer architecture delivers:
- Layer 1 — Input: AI voice typing with zero-prompt cleanup across 100+ languages
- Layer 2 — Wand: Select text anywhere, transform it by voice
- Layer 3 — Live Agents: Execute across Gmail, Slack, Notion, GitHub, WhatsApp, and 27+ apps
- Layer 4 — Autonomous Agents: Scheduled agents that run automatically — daily digests, weekly summaries, meeting prep
What's Next
Zavi is building toward a world where voice is the primary interface for all computing. Not a world of chatbots and voice assistants locked in bubbles — but a world where you speak once, and everything happens across every app you use.
Try Zavi today for free on any platform. The future of computing sounds like you.