Dashboard
Overview of your AI Data Platform
Manage Input Documents
--
Upload, manage and vectorize documents
Chat & Query
0
Ask questions across your knowledge base
Generative AI
--
Query planner, templates, multi-hop reasoning
Knowledge Graph
Not Indexed
Entity extraction, visualization, evaluation
MCP Tools
9 Tools
Model Context Protocol integration
Users
--
Manage platform users and permissions
Manage Input Documents
Upload and manage your document corpus
No Documents Loaded
Upload documents or initialize the pipeline to load from the input_docs directory.
Register Database Connection
Database Details
Manage Websites
Browse and edit website-URL CSVs under your user directory and the common input pool. Columns: WebSite, crawldepth, optional login/password for authenticated sites (leave blank for unauthenticated crawl), UpdatedOn (vectorization status — ✓ = vectorized at the timestamp shown, ○ = pending crawl), CrawlMode per row (incremental fetches only what changed, full refetches everything, off skips the row), and DownloadImages per row (off by default; enabling downloads every <img> and feeds it through the CLIP + vision-model pipeline — confirmation modal warns about cost).
No CSV Selected
Select a CSV from the list on the left to view or edit its contents.
Re-Vectorize Changed Rows?
The CSV has 0 changed row(s).
Click Yes to save the CSV and re-crawl & re-embed those rows into the publicData FAISS store. Unchanged rows keep their existing vectors.
Click No to save the CSV only.
Warning: computationally expensive
Image download will fetch every <img> tag from crawled pages, save them under <user_data>/_web_extracted_images/, run CLIP visual embedding, and call the vision model to describe each image. This is the same pipeline used for images extracted from PDFs.
Per-page and per-crawl caps apply (default 20 / 200), but API costs can still be significant for image-heavy sites.
Do you wish to proceed?
Database Connections
Register, test, and index relational and NoSQL database connections
No databases connected. Click Add Database to register a relational or NoSQL database.
Manage Saved Documents
Browse and manage your generated output files
Select a Document
Choose a file from the list to preview its contents.
Chat & Query
Generative AI
Advanced AI capabilities: query planning, templates, structured output
Multi-Step Query Planner
Decompose complex questions into sub-tasks executed as a DAG.
Structured Output Templates
Generate responses in predefined formats: reports, financial, legal, and more.
Manage Saved Documents
Click Refresh to load saved documents.
Visual Search (CLIP)
Search images by text description or find visually similar images.
Ontology Layer
Manage domain ontologies for entity and relationship extraction.
Knowledge Graph
Entity extraction, graph visualization, and quality evaluation
Index your documents to generate the knowledge graph
Graph Statistics
MINE Evaluation Metrics
Entity Details
Search Graph
MCP Tools
Model Context Protocol — integrate with Claude Desktop, Cursor, and external AI agents
rag_ingest
PipelineInitialize pipeline and vectorize documents.
rag_query
QueryExecute a natural language RAG query.
rag_plan_query
QueryMulti-step query decomposition with DAG execution.
rag_health
SystemCheck pipeline health and system status.
rag_list_sources
DocumentsList all loaded and vectorized documents.
rag_visual_search
VisualCLIP text-to-image search.
rag_find_similar_images
VisualFind visually similar images across all documents.
rag_remove_sources
DestructiveRemove specific documents from all stores.
rag_delete_all
DestructiveComplete data wipe for current user.
User Management
Manage platform users (admin only)
Loading users...
Server Administration
Graceful shutdown saves all user data (KG, indexes) before stopping the server.