| ▲ | Leftium 4 days ago | |
Epicenter[1] is attempting to go against the grain here[2]. I'm not sure if it will require one of these SaaS to become a client for this business model to work... > The long-term direction is for Epicenter to become a foundational data framework for building apps where users truly own and control their own data. In other words, the Epicenter framework becomes an SSO for AI applications. Users use Epicenter to plug into any service, and they'll own their data, choose their models, and replace siloed apps with interoperable alternatives. Developers will use our framework to build highly customized experiences for said users. To pursue that goal seriously, we also need a sustainable model that honors our commitment to open source. > ...The entire Epicenter project will be available under a copyleft license, making it completely free for anyone building open-source software. On the other hand, if a company or individual wants to incorporate the framework into their closed-source product, they can purchase a license to build with Epicenter without needing to open source their changes and abide by the copyleft terms. This is the model used by Cal.com, dub.sh, MongoDB, etc. | ||
| ▲ | DevItMan 3 days ago | parent [-] | |
Two things make local‑first frameworks break out: a boring hosted “default” and one killer app. Even if the vision is user‑owned data, ship a managed option with sane defaults so teams can try it in 5 minutes, then let them flip a switch to self‑host or bring their own storage/keys later. Pair that with a flagship app that proves the value (e.g., a shared notes/CRM/mail client where offline + conflict‑free collab is obviously better). Frameworks without a hero use case tend to stall because devs can’t justify the integration time. On the business model, dual license works if you de‑risk the integration: stable plugin ABI, permissive SDKs, and paid “closed‑source embedding” tier with SLAs and on‑prem support. Where I’ve seen revenue actually land: (1) paid sync/relay with zero data retention, (2) enterprise key management and policy controls, and (3) priority support/migration bundles. One caution: “privacy” alone doesn’t convert; solve a concrete ops pain. I built CrabClear to handle the obscure brokers others miss, and the lesson was the same—privacy sells when it eliminates a specific, recurring headache. If Epicenter can quantify one such headache and make it vanish out‑of‑the‑box, the model becomes much easier to sustain. | ||