A personal feed should be auditable
Sources, freshness, and reading depth shape the ranking.
Self-hosted RSS reader / private recommendations / no Big Tech reading profile
Personalized reading without Big Tech in the loop.
Dibao ranks only the RSS and Atom feeds you chose. It gives you the calm of RSS, the convenience of personalization, and a reading profile that stays inside your own instance.
- Sources
- Only your RSS subscriptions
- Cost
- Usable without token spend
- Profile
- Stored in your self-hosted SQLite instance
RecommendedWhy
Small embedding models for personal reading
Provider optional; baseline ranking always remains usable.
RSS can be calm without becoming dumb
No ad slots and no platform-injected sources.
Why Dibao
You should not have to choose between manipulation and overload.
Centralized feeds feel effortless, but the platform owns the source mix, the ads, and the profile. Traditional RSS respects your sources, yet collapses into an unread backlog. Dibao adds a personal ranking layer without becoming a new platform.
Comparison
Centralized platforms, classic RSS, and Dibao.
01Centralized platforms
Great flow, unclear incentives.
- Strong personalization
- Fast discovery
- Sources are platform-shaped
- Ads and ranking goals blur together
- Your profile becomes their asset
02Classic RSS readers
Clean and controllable, but easy to overload.
- No ad slots
- You choose every source
- Little personalized ranking
- Manual triage grows with every feed
- Reading behavior rarely compounds
03Dibao
RSS source ownership with a private ranking layer.
- Self-hosted data
- Recommendations stay inside subscriptions
- Explanations, diagnostics, and fallback are visible
Technical stance
AI-powered, without a running LLM bill.
Generative AI made personalized reading fashionable, but LLMs are not the cheapest engine for a constantly updating feed.
Dibao leans on recommendation algorithms and small embedding models. Run locally on a Mac or NAS, or use a free / low-cost OpenAI-compatible provider when that fits better.
LLMGood for generation, heavy for feed ranking
EmbeddingsSmall models can rank by similarity and interest
Mac / NAS / Free ProviderLocal first, cloud optional
Philosophy
Your reading profile should be an asset you own.
The things you read, save, ignore, and finish are a map of your attention. Dibao keeps that map in your own instance so it can serve your reading instead of someone else's ad system.
Behavior eventsLocal
Interest clustersExplainable
Provider settingsUser-chosen
Reading profileOwned asset
Get v0.1.0