Beever Atlas v0.1 has launched! Star us on GitHub
The LLM Wiki Way

Turn Your Data into AI Knowledge

Realtime.Effortless.Insightful.

Beever Atlas captures knowledge from your team conversations and makes it instantly searchable with AI. Open source and self-hostable.

6-Stage
AI Pipeline
<200ms
Retrieval
4+
Platforms
Apache 2.0
Open Source

Trusted by developers

Open Source
Apache 2.0 Licensed
Community
Contributors welcome
Inspired by
Karpathy's LLM Wiki

Three simple steps

How It Works

1

Connect

Link your messaging platforms in minutes. Slack, Discord, Teams, or Telegram.

2

Capture

Knowledge is automatically extracted and structured from your conversations.

3

Retrieve

Ask questions in natural language and get accurate, sourced answers instantly.

Dual-MemoryArchitecture

Two complementary memory systems work together — semantic search for meaning, knowledge graph for relationships.

Semantic Memory

Powered by Weaviate

Vector embeddings find relevant content by meaning. Handles ~80% of queries in under 200ms.

3-Tier RetrievalFuzzy SearchMultimodal

Graph Memory

Powered by Neo4j

Entity extraction and relationship mapping for multi-hop reasoning. Handles ~20% of queries.

Multi-hop TraversalTemporal ChainsEntity Links

What was discussed about JWT?

Semantic

Who decided to use RS256?

Graph

How did our auth approach evolve?

Both

FromConversationstoKnowledge

A 6-stage AI pipeline transforms scattered messages into structured, searchable knowledge.

01

Sync

Fetch messages from Slack, Discord, and Teams in real time

02

Extract

LLM extracts facts, entities, and relationships from conversations

03

Validate

Quality gates filter low-confidence extractions (≥0.5 threshold)

04

Store

Write to Weaviate (vectors) and Neo4j (graph) via outbox pattern

05

Cluster

Group related facts into topic clusters using cosine similarity

06

Wiki

Generate structured wiki pages with citations and diagrams

Output

Authentication Decisions

  • Migrated from HS256 to RS256 for asymmetric key verificationSlack
  • JWT expiry set to 15 min with refresh token rotationDiscord
  • Public keys distributed via JWKS endpoint at /.well-known/jwks.jsonSlack

Features

Everythingyourteamneeds

A complete knowledge platform, from ingestion to retrieval.

Dual Memory

Semantic vector search combined with knowledge-graph relationships for comprehensive recall.

AI Q&A

Ask questions in natural language and get grounded answers from your team knowledge base.

Multi-Platform

Works natively with Slack, Discord, and Microsoft Teams — ingest from any platform.

Knowledge Graph

Automatically extract entities and relationships to build a navigable knowledge graph.

Wiki Generation

Transform scattered conversations into structured, always-up-to-date wiki pages.

File Import

Ingest PDFs, Markdown, and documents alongside chat messages for a unified knowledge layer.

BuildwithAtlas

First-class MCP support, REST API, and Docker deployment. Get started in minutes.

mcp.json
{
  "mcpServers": {
    "beever-atlas": {
      "url": "http://localhost:8000/mcp"
    }
  }
}

Works with Claude Desktop, Claude Code, and any MCP-compatible client.

beever_askbeever_searchbeever_graphbeever_wiki

Seeitinaction

Ask a question, get answers grounded in your team's real conversations.

Beever Atlas

Knowledgethatwritesitself

Beever Atlas automatically generates structured wiki pages from your team's conversations.

Beever Atlas — Wiki

Wiki › Authentication › JWT Strategy

JWT Authentication Strategy

Updated 2 hours ago from 12 conversations


The team decided to adopt RS256-signed JWTs for the v2 API, replacing the session-based auth system. Key drivers: stateless scaling, mobile client support, and third-party integration needs.

Key Decisions

Use RS256 algorithm for JWT signing to support key rotation

Slack #engineering

Set token expiry to 15 minutes with 7-day refresh tokens

Discord #backend

Migration timeline: 3 sprints, starting Sprint 24

Teams Planning
Decision Flow Diagram

HowWeCompare

The only team-oriented LLM wiki with dual memory, auto wiki generation, and MCP support.

FeatureBeeverKhoj
Team-oriented
Dual Memory
Wiki Generation
Knowledge Graph
Multi-platform
Open Source
Chat Ingestion
Self-hosted
MCP Support

See detailed comparison →

Connectyourplatforms

Bring knowledge from every corner of your organization.

Slack

Ingest messages, threads, and files from any Slack workspace in real time.

Discord

Capture community discussions, support channels, and voice-chat transcripts.

Microsoft Teams

Connect Teams channels and chats for seamless enterprise knowledge capture.

Telegram

Import Telegram group and channel messages into your knowledge base.

More integrations coming soon

Gmail
GitHub
Notion

Use Cases

Built for every team

From product teams to open-source maintainers.

Engineering Teams

Recover architectural decisions

"Who decided to use JWT for the v2 API?"

Atlas finds the decision across Slack threads and Teams meetings — with full context on why RS256 was chosen over HS256.

Product Teams

Cluster user feedback

"What feature requests came up this sprint?"

Surface every request across Discord and Slack, clustered into themes with direct links to each source conversation.

Remote Teams

Catch up across time zones

"What happened in the incident while I was offline?"

Get a complete timeline of overnight Slack activity — decisions, owners, and action items reconstructed automatically.

Open Source Projects

Summarize community consensus

"What was the RFC consensus on the plugin API?"

Compile opinions across Discord channels into a structured summary with contributor attributions and message links.

Apr 2026

v0.1.0 — Initial Open Source Release

Dual memory, wiki generation, multi-platform sync, MCP server

Open Source

Apache 2.0 licensed. Star us on GitHub and join the community.

Ready for production?

Get enterprise-grade features for your team. Ship to production with confidence using security and compliance tools your team will approve.

  • Single Sign-On (SSO) & SAML
  • Audit logs & compliance reporting
  • Role-based access controls
  • Spend controls & usage limits
  • Dedicated support & SLA
  • Custom deployment options

Talk to the team

Tell us about your team, use case, and what you're trying to build. We read every email and typically reply within one business day.

Email hello@beever.ai

Prefer another channel? Reach us via GitHub Discussions.