Why I'm not open sourcing my Android voice agent
I built LetItDo, a voice agent for Android. WhatsApp messages, Spotify, grocery orders, the boring 47-tap routines.
Demo: https://x.com/22Gstudios/status/2051377769414791582
Every other day someone asks: are you going to open source it?
The honest answer is no.
Open source is not free distribution. Open source attracts contributors and freeloaders, not paying customers. Linear, Notion, Superhuman started closed. Cal.com later open sourced AFTER paying customers existed, as a marketing move. They did not open source to find users; they used open source to amplify reach to users they already had.
If your product has zero distribution, open sourcing gives you zero contributors and zero users with extra steps.
Open source kills monetization optionality. The moment LetItDo is on GitHub:
- Someone wraps it, hosts it, and charges for it
- Customers who would have paid me $5/month build their own from source
- Acquirers stop being interested
The "AI makes everything copyable" argument. People say: with Claude Code, anyone builds LetItDo in a week. Half right. AI compresses build time, does not eliminate moat. Distribution, brand, specific UX choices, and OEM-specific test matrix all stay valuable.
Open sourcing compresses competitor copy time from weeks to days. For an unvalidated product, worst possible time to give it away.
The general rule: Open source AFTER you have a profitable product, AS marketing. Not BEFORE you have users, AS distribution.
If you have shipped indie products and made the open source decision differently, I want to hear why.
Demo: https://x.com/22Gstudios/status/2051377769414791582
Waitlist: https://tally.so/r/jaGvx9
