wouldit.work

Consumer apps

Omnivore: a loved open-source read-it-later app that shut down when its team joined ElevenLabs

Omnivore grew to a reported 500,000 users with a free, open-source reader. In October 2024 the team joined ElevenLabs and the hosted service closed within weeks. A lesson about free products and the cost of running them.

Product
Omnivore
Founder
Jackson Harper and Hongbo Wu
Launched
2021
Model
Free app; team joined ElevenLabs

Published · 2 min read · 4 sources

Key takeaways

  • Launched in 2021 for 'people who like text'; ElevenLabs said the app was used by 500,000 people when the team joined in October 2024.
  • Users got only a few weeks to export before their data was deleted; the hosted service shut down.
  • The code stays open source on GitHub, but most users wanted the hosted app, not a codebase.

What it is

Omnivore was an open-source read-it-later app with highlighting, PDF and offline support, apps for web, iOS and Android, browser extensions, and text-to-speech powered by ElevenLabs voices (TechCrunch).

Timeline

  • 2021: Jackson Harper and Hongbo Wu launched Omnivore as a read-it-later app for "people who like text" (TechCrunch).
  • Before the deal: Omnivore integrated ElevenLabs' AI voices; listening became "one of our most popular features", the founders wrote (TechCrunch).
  • 29 Oct 2024: ElevenLabs announced that the Omnivore team, whose app it said was "used by 500,000 globally", had joined to work on its ElevenReader app (ElevenLabs).
  • 16 Nov 2024: deadline to export data, after which it would be deleted (TechCrunch).

Why it worked

  • A focused audience. It was built for serious readers, with features like highlights and PDFs (TechCrunch).
  • Free and open source, and good enough to recommend. The Verge's David Pierce said he had recommended it "a lot"; his own paid pick, Readwise Reader, he called "too expensive" (The Verge).
  • A partner integration that became the exit. Using ElevenLabs' API for text-to-speech led directly to the team joining ElevenLabs (ElevenLabs).

What to learn

  1. A big user count is not a business model. A popular free product still has server bills, and someone has to pay them.
  2. If people store their data in your product, plan a shutdown path (exports, timelines) before you need it.
  3. Your best feature might be a partner's tech. That is useful leverage, and also a sign of where an acquirer's interest lies.

Risks and what didn't work

  • Users had only about two weeks to export their data (The Verge).
  • ElevenReader is a text-to-speech reader, not a replacement read-it-later app, so users had to migrate elsewhere (TechCrunch).
  • Open-sourcing the code keeps it alive for self-hosters (GitHub), but that only helps the technical minority.
  • The 500,000 user figure comes from ElevenLabs' announcement and is not independently verified (ElevenLabs).

FAQ

Short answers drawn from the sections above.

What is Omnivore?

Omnivore was an open-source read-it-later app with highlighting, PDF and offline support, apps for web, iOS and Android, browser extensions, and text-to-speech powered by ElevenLabs voices (TechCrunch).

Why did Omnivore work?

  • A focused audience. It was built for serious readers, with features like highlights and PDFs (TechCrunch).
  • Free and open source, and good enough to recommend. The Verge's David Pierce said he had recommended it "a lot"; his own paid pick, Readwise Reader, he called "too expensive" (The Verge).
  • A partner integration that became the exit. Using ElevenLabs' API for text-to-speech led directly to the team joining ElevenLabs (ElevenLabs).

What can indie builders learn from Omnivore?

  1. A big user count is not a business model. A popular free product still has server bills, and someone has to pay them.
  2. If people store their data in your product, plan a shutdown path (exports, timelines) before you need it.
  3. Your best feature might be a partner's tech. That is useful leverage, and also a sign of where an acquirer's interest lies.

What didn't work for Omnivore, and what are the risks?

  • Users had only about two weeks to export their data (The Verge).
  • ElevenReader is a text-to-speech reader, not a replacement read-it-later app, so users had to migrate elsewhere (TechCrunch).
  • Open-sourcing the code keeps it alive for self-hosters (GitHub), but that only helps the technical minority.
  • The 500,000 user figure comes from ElevenLabs' announcement and is not independently verified (ElevenLabs).

Sources

Every link cited above, in order of first use.

  1. 1.TechCrunch (techcrunch.com)
  2. 2.ElevenLabs (elevenlabs.io)
  3. 3.The Verge (theverge.com)
  4. 4.GitHub (github.com)

Related case studies

More in Consumer apps →