My app failed, then I changed one thing, and made $80K

Starter Story 12min #97
My app failed, then I changed one thing, and made $80K
Watch on YouTube

Summary

  • Praneeth

    • Quit his $250,000 software engineering job to go all in on building his own apps
    • Built an app originally called Rabbit Holes AI, later renamed Canvas Mode
    • The first version of the app launched with a freemium subscription model and saw almost no traction
    • Pivoted the pricing model and rebuilt the app as a one-time purchase desktop application
    • After the pivot, the app generated $80,000 in revenue over the following 6 months
    • Currently has around 1,200 paying users and approximately 38,000 unique site visitors
  • Products and Offerings

    • Canvas Mode (formerly Rabbit Holes AI): A canvas-first AI interaction app designed for power AI users
      • Allows users to branch conversations into nodes instead of relying on linear chat
      • Built as a desktop app using Electron so it runs entirely on the user’s machine
      • Uses a bring-your-own-keys (BYOK) model where users supply their own AI API keys
    • Started as a web app with a free trial and subscription, then pivoted to a one-time purchase lifetime deal
  • Metrics and Financials

    • Gross revenue: Approximately $80,000
    • Profit: Around $75,000 after refunds
    • Paying users: About 1,200
      • 60% are lifetime deal users
      • 40% are one-time purchase users who must renew after a year for continued updates
    • Monthly operating costs: Roughly $250
      • $50 for landing page hosting on Vercel
      • $200 for Cursor subscription (for ongoing feature development, not maintenance)
  • Strategy and Growth

    • Pivoted from a freemium subscription to a one-time purchase lifetime deal to reduce his own costs and attract more committed users
    • Offloaded AI API costs and server costs entirely to users by switching to a desktop app with BYOK
    • Started pricing at $29 with under 100 users, raised to $49 after crossing 100 users, and plans to move toward a recurring model as the user base grows
    • Growth tactics:
      • Tailored landing page copy specifically to power AI users, calling them out directly in the H1 and throughout the site
      • Engaged on Twitter by replying to conversations about AI interactions rather than directly pitching the product
      • Focused on getting the ideal customer profile (ICP) right before scaling outreach
  • Tech Stack and Infrastructure

    • Landing page: Built with Next.js, hosted on Vercel
    • Desktop app: Built with Electron, using React and TypeScript under the hood
    • Hosting: Cloudflare
    • AI integration: Uses Vercel’s AI SDK for abstraction across multiple AI providers, eliminating the need to manage individual API integrations
    • Development: Used Cursor as an AI coding assistant throughout the build process
  • Lessons and Advice

    • Recommends lifetime deals for nearly every bootstrap business, even if a subscription model already exists, because it provides a cash boost, validates the idea, and attracts users who give better feedback
    • Advises against starting with a freemium subscription model when bootstrapping, since VC-style growth tactics delay finding product-market fit
    • Emphasizes that founders should focus on building something people will pay for rather than appealing to peers or building in isolation
    • Stresses the importance of finding product-market fit as quickly as possible and letting market validation, not personal assumptions, guide what to build
    • Recommends the bring-your-own-keys model for AI products to eliminate API costs and make the business nearly zero-cost to operate
Back to Starter Story