Quivr
Go engine that ingests news, mail and social feeds, indexes them for hybrid search and sends topic alerts.
About
Quivr has been rebuilt from the ground up. The old second-brain chat app is gone; the repository, now under The Vibe Company, holds Quivr V2, a Go service for ongoing monitoring: it pulls in news feeds, mailboxes or social posts as they appear, indexes each item almost at once, and alerts you when new material fits a topic you follow. One binary runs the API, worker and migrations, backed by PostgreSQL, S3-compatible object storage, Temporal for durable workflows, Weaviate for lexical, vector and hybrid queries, and a local server running a small multilingual E5 embedding model. Writes are idempotent and versioned, so corrections keep history. Saved queries become subscriptions that fire signed webhooks on keyword rules, local embedding similarity, or an optional third-party classifier that receives article text. Connectors cover RSS, Microsoft 365 mail, X lists and archive imports, and plugins add formats and business rules. A local stack needs Docker Compose, Go, Python and Node on Linux x86_64 or Apple Silicon. The API is still v0 and releases are alphas the team says are not production ready. Licensed MIT.
Should you use Quivr?
Pick it when
Consider Quivr when you are building a personal second-brain style assistant over your own files and want to evaluate a ready RAG engine that works with Anthropic, OpenAI, Mistral, or local Ollama models.
Look elsewhere when
Skip it for anything production-bound: the repository README marks it as evaluation stage with a v0 API that may change without notice. AnythingLLM or Khoj are usable second-brain apps, and LlamaIndex suits custom builds.
Alternatives to Quivr
- Khoj
Personal second brain with Obsidian, Emacs, phone, and WhatsApp clients plus scheduled automations, but AGPL-3.0 licensed.
- AnythingLLM
MIT desktop and Docker app with workspaces and many model providers, usable for chatting with documents today without writing code.
- LlamaIndex
MIT framework with loaders and query engines, more code to write but a widely integrated foundation for a custom document assistant.
- Kotaemon
Apache-2.0 document QA web UI with hybrid retrieval that developers can extend, a working app rather than an engine still in evaluation.
Reviews (0)
Leave a Review
No reviews yet. Be the first to review!
Details
- Category
- RAG & Document Retrieval
- Price
- Free
- Platform
- Local/Desktop
- Difficulty
- Easy (2/5)
- License
- MIT
- Added
- Apr 3, 2026