Documentation
Everything you need to get KnownBase running for your team.
Quickstart (5 minutes)
Step 1: Add KnownBase to Slack
Click "Add to Slack" and authorize KnownBase. You'll select which channels KnownBase can access.
channels:history # Read channel messages
channels:read # List channels
users:read # Read user profiles
chat:write # Respond when asked
Step 2: Select channels
Choose which channels KnownBase should learn from. We recommend starting with 3-5 high-activity engineering channels. KnownBase only accesses channels you explicitly approve.
Step 3: Ask your first question
After 24 hours of learning, try asking:
@knownbase What did we decide about API versioning?
@knownbase Show me the expertise map for the backend team
How KnownBase Works
The People Graph
KnownBase builds a knowledge graph (Neo4j) from your team's conversations. For each person, it learns:
- • Expertise areas — what topics they discuss and how deeply
- • Communication style — formal, casual, technical depth
- • Relationships — who helps whom, who collaborates frequently
- • Decisions — what they decided, when, and why (linked to source)
- • Evolution — how their expertise changes over time (temporal versioning)
4-Tier Memory Architecture
KnownBase uses a hierarchical memory system to never forget:
- • Tier 1 — Working Memory: Last 20 messages, verbatim
- • Tier 2 — Rolling Summaries: Compressed history, refreshed every 10 messages
- • Tier 3 — Semantic Search: Vector embeddings for historical recall
- • Tier 4 — Knowledge Graph: Structured facts with temporal versioning
Anti-Hallucination (CoVe Pipeline)
Every factual answer goes through a 4-stage Chain-of-Verification:
- 1. Classify — Is this a factual question?
- 2. Extract claims — Break response into atomic facts
- 3. Verify — Check each claim against sources
- 4. Reconstruct — Rebuild response with confidence levels and citations
Integrations
💬 Communication
Slack (native) · Microsoft Teams · Discord · Email (Outlook/Gmail)
💻 Development
GitHub · GitLab · Bitbucket (via MCP)
📋 Project Management
Jira · Linear · Asana · Notion (via MCP)
🎥 Meetings
Zoom · Google Meet · MS Teams meetings · Circleback · Fireflies
Additional integrations available via MCP (Model Context Protocol). Custom integrations on Enterprise tier.
API Reference (Coming Soon)
Query KnownBase's People Graph and Decision Memory programmatically. Available on Business and Enterprise tiers.
GET /api/v1/expertise?topic=payments
# Get decision history
GET /api/v1/decisions?query=why+kafka
# Get person profile
GET /api/v1/people/{user_id}/profile
Full OpenAPI spec will be available at launch. Join waitlist for early API access.