Inspiration

Everyone has something they want to say but can't — to a crush, a friend, a colleague. Existing anonymous messaging apps feel either too sketchy or too cold. I wanted to build something that felt warm, real-time, and safe. But beyond just building it, I wanted to actually understand how people use it from day one. That's where Novus.ai came in — I didn't want to ship into the void.


What it does

Whspr gives you a personal anonymous chat link. Share it anywhere — your Instagram bio, WhatsApp, Twitter — and anyone can send you a message without revealing who they are. You reply in real-time from your dashboard. Every conversation auto-deletes after 24 hours, leaving no trace. It's like NGL meets a real chat experience.

With Novus.ai installed, I can see exactly how users move through the app — where they drop off, which pages they land on, how long they stay in a conversation, and whether they actually share their link after signing up. That data shapes every iteration.


How we built it

  • Frontend: React + Vite + TypeScript + Tailwind CSS v4
  • Backend: Supabase (Auth, Postgres, Realtime subscriptions)
  • Routing: React Router v6
  • Deployment: Vercel
  • AI tooling: Lovable as the primary build agent, Claude for debugging and architecture
  • User Analytics: Novus.ai — installed from day one to track real user behavior, session flows, and feature engagement

Novus.ai gave me visibility I wouldn't have had otherwise. Instead of guessing whether users were actually sharing their links or dropping off at the auth page, I could see it in real time and act on it.


Judges Section

Test Account Username: test1 Password: test1234

Challenges we ran into

The biggest challenge was deployment. The project started on TanStack Start (SSR), which Lovable uses internally but doesn't play well with standard hosts like Vercel. I spent significant time debugging Nitro presets and build pipelines before migrating to plain React + Vite — which immediately unblocked everything.

Other challenges included Supabase RLS policy mismatches, realtime subscription stale closures, and getting Open Graph previews working for WhatsApp sharing.

Novus.ai helped surface one issue early — I could see users landing on the public chat page and leaving without sending a message. That pointed me toward a UX problem with the "Start chatting" flow that I was able to fix quickly.


Accomplishments that we're proud of

  • Shipped a fully working real-time anonymous messaging app from scratch
  • Integrated Novus.ai from day one, giving real insight into how users actually behave
  • Migrated from a complex SSR setup to a clean SPA without losing any features
  • Swipe-to-reply gesture on mobile that feels native
  • WhatsApp-ready share links with OG image previews
  • Clean, polished UI that feels genuinely fun to use

What we learned

AI coding agents are powerful but make opinionated infrastructure choices that create deployment headaches. Validate your deployment target early.

The bigger lesson was about shipping with insight, not just shipping. Having Novus.ai installed from the start meant I wasn't flying blind. I could see real people using the product in real time — which pages they hit, where they bounced, what they actually did. That feedback loop is the difference between shipping and shipping something that actually works for people.


What's next for Whspr

Novus.ai data will directly drive the next roadmap. Early signals I'm already watching:

  • Drop-off on the auth page → simplify the signup flow
  • Low link-share rate after signup → improve the dashboard onboarding moment
  • Conversation depth → are users having real back-and-forth conversations or just receiving one message?

Features in the pipeline:

  • 24-hour auto-deletion via Supabase Edge Functions
  • Push notifications for new messages
  • Reaction support on messages
  • Content moderation layer
  • Mobile app (React Native)

Built With

Share this project:

Updates

Submission history