Choosing transparency and self-hosting over a hosted engine

For teams that need to own the model, the engines, and the editorial layer, slorg replaces an opaque hosted answer engine with six functions they can read and change.

The problem

Hosted answer engines own the system prompt, the model choice, the retrieval backends, and the ranking logic. For a team with data-residency, cost, or auditability constraints, that lack of control is a non-starter — and the behaviour can change under them without notice.

With slorg

slorg is MIT-licensed and self-hosted. You bring your own OpenAI-compatible key, pick your engines via SearxNG, choose your model, and read the entire pipeline in the source. Nothing about the planning is hidden or subject to a vendor’s roadmap.

How it flows

Self-host — Run slorg and your own SearxNG instance on your infrastructure.
Configure — Set the model, engines, and result limit through a handful of env vars.
Inspect — Read the six-step source to understand exactly what runs.
Extend — Fork or wrap it — the editorial layer is yours to change.

What you get

Full control over model, engines, and ranking
Behaviour is stable and inspectable
No vendor lock-in or silent changes

try it Install with npm install lorg and follow the quickstart, or read how the pipeline works on the architecture page.