Connect without surrendering structure
Use the internal wiki and external systems as parallel sources instead of forcing another migration.
The governed knowledge layer
One infrastructure for company knowledge and every AI.
Create knowledge directly, connect the systems teams already use, and deliver explainable evidence through search, MCP, and API without detaching source permissions.
01Wiki + connectors
02Identity + ACLs
03Evidence Fusion
04Graph context
05MCP + agents
06Incremental sync
Platform principles
The control boundary stays consistent.
Every interface uses the same document identity, permission decisions, provenance, and incremental update model.
Resolve access before retrieval
Identity, roles, workspace scope, and source ACLs stay in the data path from search to download.
Return an evidence contract
Every result can carry a stable file ID, original link, citations, retrieval signals, and related paths.
Improve knowledge deliberately
Authorized people and agents can propose versioned updates without bypassing review or provenance.
One product state
Four surfaces for the complete knowledge lifecycle.
People author and inspect knowledge, teams search it, agents access it, and operators keep it healthy without switching between disconnected control models.
Knowledge Base
Create durable company knowledge, inspect originals, maintain notes, and work across private and shared workspaces.
- Linked workspaces
- Native file previews
- Versioned notes
Discover
Search authorized knowledge, inspect why an item matched, and continue through source files and related context.
- Explainable results
- Graph exploration
- Original access
Agent Access
Issue scoped credentials for MCP and REST while keeping user identity, permissions, and audit context attached.
- OAuth and API keys
- Scoped tools
- Revocable access
Operations
See connector health, failed documents, queue lag, audit events, repair actions, and recovery status in one place.
- Diagnostics
- Repair workflows
- Backup and restore
Request lifecycle
A query never jumps straight to an index.
Every request establishes identity and boundaries before gathering evidence, then verifies candidates again before anything leaves the platform.
Search request
quarterly renewal termsProfilebalanced
Calleruser_04f2
Resolve identity
Establish the user, role, tenant, and credential scope.
Build boundaries
Translate workspace and source permissions into retrieval filters.
Gather evidence
Run the selected lexical, semantic, metadata, and graph channels.
Verify candidates
Recheck every result against current permissions and source state.
Return a contract
Deliver evidence with IDs, citations, reasons, and original actions.
Shared evidence contract
Web, REST, and MCP receive the same truth.
Each interface is optimized for its caller, but stable file identity, source provenance, access decisions, and traversal boundaries remain consistent.
Choose an interface
WebHuman inspection and source preview
REST APIApplication and workflow integration
MCPAgent search, retrieval, and traversal
COGLAKEEvidence contract
- file_id
- file_f84c2
- source
- Google Drive / Sales
- why
- semantic + exact phrase
- access
- verified
- original
- available
- relations
- 3 bounded paths
The payload changes shape for each interface. Its identity, access decision, and provenance do not.
Day-two operations
Know when knowledge is current, delayed, or broken.
Operational signals connect sync, processing, indexing, permissions, and recovery so teams can diagnose and repair the actual failing stage.
Operations healthAll critical paths visible
LiveSourceEventStatusDetail
Google Drive / SalesDelta syncCompleted14 changed, 2 deleted
Internal / ProductDocument processingRequeuedPresentation extraction
PermissionsACL repairVerified0 drifted documents
RecoveryEncrypted snapshotCompletedRestore manifest verified
Knowledge is the foundation. AI Workspace is the next step.
Use the Knowledgebase independently through search, wiki, API and MCP. Add AI Workspace for integrated chats, personal agents and automations.
Start with your real knowledge
Turn company knowledge into governed context.
See how CogLake fits your sources, permissions, and AI interfaces.
Book a walkthrough