JMP Expert System

Statistical analysis, scripting, and data workflows — grounded in real JMP documentation.

An AI-powered expert system backed by 20 indexed JMP documentation PDFs. Ask questions about statistics, DOE, SPC, JSL scripting, Python integration, or data access and get context-grounded answers with source citations. You will always get JSL when you ask for a worked solution, and if JMP is available through the MCP execution path it can be opened in JMP automatically.

Open the expert system View reference library
Primary support areas Statistics, JSL, Python integration, and data access
Operating pattern Clarify the problem, reproduce it, then teach through the fix
Domain base DOE, SPC, capability analysis, distribution fitting, and manufacturing troubleshooting

Expert system

Ask questions about statistics, DOE, SPC, JSL scripting, Python integration, and data workflows. The system retrieves relevant context from indexed JMP documentation and generates a grounded answer with source citations.

Tip: Ask for a worked example and you will get JSL. Use Show in JMP to route that script through the MCP execution path; if JMP is available there it will open in JMP, otherwise you still get the JSL with a saved-script fallback.

Try JMP Expert before you sign in.

Ask about statistics, DOE, SPC, JSL scripting, Python integration, or data workflows. Demo answers use the same source-grounded retrieval path as the signed-in workspace.

Start a temporary demo chat without creating an account.
Sign in to keep chat history and manage uploaded reference PDFs.
Create an account when you want a persistent JMP support workspace.

Choose how to enter

Use the demo for a quick question, or sign in when the work should persist.

Sign in Create account
Document uploads and saved sessions require an account.

Document library

Upload JMP documentation, guides, and reference PDFs. Documents are automatically indexed for the expert system's retrieval pipeline.

Login required

Document management requires authentication. Log in to upload and view documents.

How the RAG pipeline works

  • Upload a PDF — text is extracted with PyMuPDF and chunked into ~500-token segments.
  • Each chunk is embedded using OpenAI text-embedding-3-small and stored in PostgreSQL with a pgvector HNSW index for fast similarity search.
  • When you ask a question, the system embeds your query and finds the most relevant chunks via indexed vector similarity (pgvector HNSW), then sends them as context to GPT-4o.
  • The answer streams in real-time and includes source citations.

Integration details

API base URL

/api/jmp-expert/

Endpoints

POST /chat/ · GET /documents/ · POST /documents/upload/ · GET /documents/<id>/pdf/