π Complete Tutorial: Configuring a Medical Chatbot with AI Engine
*This is the fifth and final part of the complete configuration guide. If you missed the first four parts, you can find them [here]*, [here]**, [here]*** and [here]****.
In the first four parts we covered:
β
Technical Foundation – Token system, cost optimization, Dashboard modules
β
Chatbot Configuration – AI Model, Embeddings, Thresholds, Knowledge Base
β
User Interface – Appearance, Popup, UI Builder, Advanced
β
Advanced Integrations – Cross-Site, Shortcodes, Actions
β
Monitoring – Discussions, Insights
β
Complete Global Settings – Chatbot, Orchestration, Files & Media, Remote Access, Others
—
π―NOW WE WILL ANALYZE THE KNOWLEDGE BASE SETTINGS
In this final part, we will complete the last piece of the puzzle:
πΉ Knowledge – Centralized management of knowledge bases
- Configuration and synchronization of medical databases
- Optimizations for efficient semantic search
- Maintenance and update strategies
βοΈ KNOWLEDGE CONFIGURATION – KNOWLEDGE BASES
1. “KNOWLEDGE” TAB – DATABASE MANAGEMENT
π AI Engine β Knowledge
π― ENVIRONMENT CONFIGURED IN SETTINGS β KNOWLEDGE:
Environment: Quadrant (β Selected in AI Engine β Settings β Knowledge) Status: Active and synchronized Configuration: Complete settings for the medical vector base
πΉ PANEL 1 – “EMBEDDINGS” – DATABASE STATUS
π― ACTIVE MEDICAL DATABASE:
Title: "Overview of Prescription Medications and Their Uses" (β Active) Model: "Embedding Ada-002, 1536 dimensions" Status: Visual duplicate in the interface (technical display) Dimensions: 1536 vectors - optimal for medical complexity
π§ AVAILABLE ACTIONS:
Refresh: Refreshes status and synchronization Useful: Marks useful content for optimizations Ret: Retention and storage management options
πΉ PANEL 2 – “MODE” – SEARCH MODES
π― QUERY CONFIGURATION:
Query Mode: (β Selected by default) - Standard search optimized for users - Ideal for direct medical interactions Expert Mode: (β Unconfigured) - Advanced search with custom parameters - For complex specialized medical cases
πΉ PANEL 3 – “ENVIRONMENT” – ENVIRONMENT CONFIGURATION
π― ADVANCED TECHNICAL SETTINGS:
Info: Technical details about the current environment Sync: Synchronization status with the database
βοΈ “SYNC” SUBTAB – SYNCHRONIZATION CONFIGURATION:
Enable Sync: [ ] Disabled (β RECOMMENDED) Function: "Embeddings will be created, updated or removed based on posts activity" Medical Motivation: Β - Our drug database is static and controlled Β - We do not want automatic synchronization with WordPress posts Β - We maintain a clear separation between medical content and the blog
π “INFO” SUBTAB – PERFORMANCE PARAMETERS:
Min Score (%): 70 (β Optimal) Max Results: 5 (β Ideal for medical) Function: "The best 5 embeddings with a score of 70% or more will provide context" Medical Motivation: Β - 70% ensures high medical relevance Β - 5 results provide optimal coverage without overload Β - Perfect balance between precision and performance
πΉ PANEL 4 – “BUILD KNOWLEDGE” – KNOWLEDGE BASE CONSTRUCTION
π― MANAGEMENT OPERATIONS:
Create New: [+] Button for new databases Push All (1): [β] ENABLED - Synchronizes all content Posts: [ ] Option for post synchronization Upload PDF: [ ] Disabled - β WE DO NOT use PDFs
π‘ IMPLEMENTATION STRATEGY:
Push All Active: Β - Ensures that all medical knowledge is synchronized Β - The database is always current and complete Β - Zero risk of missing medical information Posts Disabled: Β - Clear separation: medical base vs blog content Β - Prevents "contamination" with non-medical information Β - Maintains an exclusive focus on medicine Upload PDF Disabled: Β - Our base is structured (JSON/database) Β - PDFs introduce variability and parsing issues Β - Total control over data format and quality
π― FINAL CONCLUSION OF KNOWLEDGE TAB:
The Knowledge system is configured for:
β Maximum performance (70% score, 5 results, Ada-002)
β Medical security (sync disabled, clear separation)
β Controlled management (only our medical database)
β Scalability (ready for additional bases)