Chidi - AI Learning Assistant

Model Details

Model Name: Chidi

Version: 2.0

Model Type: Multi-Agent LangGraph System

Developed By: Sand Technologies

Model Description: Chidi is an advanced multi-agent AI system built with LangGraph that provides intelligent tutoring and support. The system features specialized agents for customer support and learning assistance, with RAG-powered document retrieval and platform-specific personalization.

How to use it: Click on the chat icon in the lower right corner to start a conversation with Chidi. The assistant adapts its behavior based on the platform you're using - providing course support on LMS platforms, administrative help on eHub, and general information on the front page.

What Chidi Can Do

Platform-Specific Capabilities

🎓 LMS Platforms (Savanna & Athena): Full access to both learning support and customer support. Can help with course content, assignments, and technical issues.

🛟 eHub Support Portal: Focused on customer support, account management, and issue resolution. Direct access to ticketing system.

🌍 ALX Africa Front Page: General information, program overview, and course selection assistance. Helps prospective students choose the right path.

Intended Use

Primary Use Cases:

• Assisting students with understanding and applying concepts in the AiCE course.
• Providing support for navigating the LMS, Savanna, Nile, and ALX.
• Answering general questions related to course logistics and student resources.

Potential Applications:

• Educational support for students enrolled in any ALX courses.
• Customer support for issues related to LMS and other applications.
• General information dissemination about the course and related platforms.
• Automation of Customer Support workflows.

Architecture & Technology

Core Technology: LangGraph v2 with PostgreSQL state persistence and Qdrant vector database for RAG

Multi-Agent System: Intelligent routing between specialized agents:

• Main Assistant: Entry point that routes to appropriate specialized agents
• Learning Support Agent: Handles educational content questions with course-specific RAG
• Customer Support Agent: Manages logistics, account issues, and platform navigation

Platform Personalization: Dynamic theming and personas based on platform context (Savanna, Athena, eHub, AlxAfrica)

Real-time Features: Streaming responses with thinking indicators during tool usage

Performance expectation

Current Capabilities:

• Intelligent Routing: Automatically determines the best agent for each query
• Context-Aware RAG: Retrieves relevant documents from course-specific collections
• State Persistence: Maintains conversation history across sessions
• Platform Integration: JWT-based authentication with platform context
• Response Time: Typical response within 3-7 seconds

New Features:

• Multi-round message formatting with visual separators
• Dynamic theme loading (80% CSS size reduction)
• Platform-specific conversation starters
• MCP tool integration for extended capabilities

Limitations

Unreliability: As with all LLMs, there is a possibility of generating incorrect or irrelevant information. The model may not always understand the user's needs accurately.

Incomplete Information: The model may not have access to the most up-to-date information or specific details required by the user.

Limited teaching scope: This model currently only knows about the existence of the learning content of AiCE.

Known Issues:

• URLs generated by the model might occasionally miss a character, rendering them non-functional.
• Functionalities like "ToCustomerSupport" and "ToLearningSupport" may not always work as intended due to limitations in context understanding.
• The chat application does not scroll with page. This is not the production environment behaviour.

Privacy:

• Conversations are stored to improve our product, but no identifying data is included.
• Compliant with data protection regulations, including GDPR.

Usage Guidelines

Target Audience: Students enrolled in the AiCE course and users of the LMS, Savanna, Applications, and ALX.

Usage Instructions:

• For Educational Support: Ask specific questions related to course content like:
  • "What is GenAI?"
  • "How can I create a CV with AI?"
  • "How will AI affect my job?"
• For Navigation Assistance: Describe the issue or the information you need about the LMS or other applications. For example:
  • "How do I log into Savanna?"
  • "How do I submit my milestone?"
• For General Inquiries: Provide concise questions for general queries about ALX, its programs, and its courses, for example:
  • "What is ALX?"
  • "When was ALX started?"
  • "What is Karibu?"

Disclaimer: This model is a tool to assist and support users. It is not a substitute for professional advice. Users should exercise caution and verify critical information independently.