--
Cognee Datasets
Each dataset is a collection of documents uploaded to Cognee. Data from the Data Lake is grouped by tab name (e.g., "client_isp_details", "transcripts"). Each dataset can be cognified independently to build the knowledge graph.
--
Graph Nodes
Nodes are entities extracted by AI from your data: people (clients, DSPs), places (houses), medical conditions, dates, and document chunks. Each node is a distinct concept in the knowledge graph that can be queried.
--
Graph Edges
Edges are relationships between nodes, e.g. "Tyler Bond WORKS_AT Casa Aurora" or "Test Patient HAS_CONDITION Epilepsy". More data produces a richer relationship network for answering complex queries.
--
Memobase Users
Each user represents a tracked entity: a client, DSP (employee), or house. Memobase builds a structured profile for each user by extracting information from every piece of data sent about them via the Data Lake.
--
Profile Entries
Each entry is a specific fact about a user, organized by topic and sub-topic. Example: health_status > diagnoses = "Epilepsy, Diabetes". More entries mean a more complete profile that AI can use for personalized responses.
--
Connection Status
Whether this dashboard can reach both Cognee and Memobase APIs. Both run on the Contabo VPS accessed via HTTPS through Traefik. Green dots in the header show individual service status.
Activity Log
Click "Connect" to load data from Cognee and Memobase.
Knowledge Graph
Entity
Entity Type
Document Chunk
Text Document
Text Summary
Extracted Entities
Connect and load the graph to see entities.
Query the Knowledge Graph
How search types work
Graph Completion uses knowledge graph relationships to answer questions like "Who works at X?" or "What diagnosis does Y have?". Text Chunks searches the raw document text for relevant passages, best for finding specific details from the original data.
Enter a query above and click Search.
Registered Users
What are users?
Each user is an entity (client, DSP, or house) tracked by Memobase. Click a user to see their full profile built automatically by AI from all data sent about them through the Data Lake pipeline.
Connect to load users
Structured Profile
How profiles work
AI reads all data about this person and extracts structured information into topics (health_status, risk_factors, certifications, etc.) and sub-topics. Each card below is one extracted fact. Profiles update automatically as new data arrives.
Select a user above to see their profile
Context for AI Prompts
What is the context?
Pre-formatted text that gets injected into an AI system prompt so it "knows" this person. Contains the profile summary and recent events. Copy this into any LLM conversation for full knowledge about this entity.
Select a user to see their AI-ready context.