# Nolvin

> Nolvin is agentic knowledge graph infrastructure. Describe a knowledge domain in one sentence; agents write a brief, discover and crawl sources on the open web, and maintain a POLE knowledge graph (Person, Object, Location, Event) with provenance — source page and verbatim quote — on every fact. The graph is queried through a REST API (91 endpoints) with graph traversal, search, timeline and GeoJSON map projections, webhooks and share links.

Key facts:
- Entity kinds: person (role, aliases), object (organizations, products, programmes; category, identifier), location (lat, lng, address), event (starts_at, ends_at). Up to six custom attributes per domain (e.g. email, phone), kept only when found verbatim on the page.
- 15 canonical relations: works_at, leads, founded, member_of, part_of, located_at, organizes, teaches, participates_in, invested_in, partner_of, owns, operates, offers, related_to.
- Re-reads are incremental: unchanged pages cost no model call, changed pages are read as a diff.
- Crawler: NolvinBot, honors robots.txt and Crawl-delay.
- API base: https://api.nolvin.com, Bearer API keys (nlv_…), read or full scope.

## Docs

- [API documentation](https://nolvin.com/docs/): Nolvin API docs: build a knowledge graph from one sentence, query entities, relationships, timelines and maps, and subscribe to webhooks. REST and OpenAPI.
- [Quickstart](https://nolvin.com/docs/quickstart): Create an API key, start an autopilot domain from one sentence, follow its progress, and query entities, timelines and maps — in five requests.
- [Authentication](https://nolvin.com/docs/authentication): Authenticate with project API keys, choose read or full scope, rotate and revoke keys, understand member roles, and publish read-only share links.
- [Concepts](https://nolvin.com/docs/concepts): Projects and domains, POLE entities and custom attributes, the gate, pages and crawl runs, autopilot stages, provenance, freshness and spend caps.
- [Querying the graph](https://nolvin.com/docs/graph-queries): Read a Nolvin knowledge graph: filter entities, open one with its sources, traverse neighborhoods, search, and project any entity onto a timeline or map.
- [Webhooks and events](https://nolvin.com/docs/webhooks): Signed webhooks for crawl.finished and budget.hit, verifying X-Nolvin-Signature, retries, and live progress through event polling or server-sent events.
- [API reference](https://nolvin.com/docs/api-reference): All 91 endpoints of the Nolvin REST API v1 — domains, autopilot, crawling, graph queries, curation, sharing, webhooks — with OpenAPI 3.1.
- [OpenAPI 3.1 spec](https://nolvin.com/openapi.json): machine-readable description of every endpoint

## Product

- [How it works](https://nolvin.com/how-it-works): From one sentence to a maintained knowledge graph: brief, source discovery, polite crawling, sourced extraction, self-assessment and diff-based re-reads.
- [The POLE model](https://nolvin.com/pole-model): The POLE model — Person, Object, Location, Event — as Nolvin uses it: per-kind facets, 15 canonical relations, provenance, timeline and map projections.
- [Use cases](https://nolvin.com/use-cases): Four live knowledge graphs, each from one sentence: a self-updating event calendar, a regional industry map, an investor network and a cultural scene.
- [About Nolvin](https://nolvin.com/about): Knowledge about people, organizations, places and events is scattered across small websites and goes stale. Nolvin turns it into sourced, living graphs.

## Optional

- [NolvinBot](https://nolvin.com/bot): NolvinBot is the crawler behind Nolvin: how it identifies itself, honors robots.txt and Crawl-delay, how often it visits, and how to limit or block it.