PBA PL Demo - AI-Powered Student Project Support Bot¶

An intelligent Telegram bot that helps students discover and learn about academic projects across different terms, senior years, and subjects. The bot uses AI to provide contextual responses about project titles and descriptions, with collaborative prompt management through PromptLayer and comprehensive conversation logging.

Purpose¶

No business context provided yet — add a context.md to enrich this documentation.

Based on the workflow implementation, this bot appears to serve students and educators by: - Providing instant access to project information across different academic terms and subjects - Maintaining conversation context to deliver relevant, personalized responses - Enabling non-technical teams to collaboratively manage and update bot behavior through PromptLayer - Tracking all interactions for analysis and improvement

How It Works¶

Message Reception: A user sends a message to the Telegram bot
Prompt Retrieval: The system fetches the latest approved prompt from PromptLayer using the "prod" tag
AI Processing: The Support Bot agent processes the user's message using:
- The retrieved prompt as system instructions
- Conversation history from memory
- Access to project database through Airtable tool
Response Generation: The AI model (Google Gemini) generates a contextual response
Message Delivery: The bot sends the response back to the user via Telegram
Interaction Logging: The complete conversation is logged to PromptLayer for analysis

The bot can search for specific projects by term, senior year, and subject, providing students with detailed project descriptions and challenges they aim to solve.

Workflow Diagram¶

graph TD
    A[Message to bot] --> B[Pull Prompt from PromptLayer]
    B --> C[Support Bot]
    D[OpenRouter Chat Model] --> C
    E[Simple Memory] --> C
    F[Get project titles and descriptions] --> C
    C --> G[Send Bot's Response]
    G --> H[HTTP Request - Log to PromptLayer]
    I[Webhook] 
    J[Embeddings Google Gemini]

Trigger¶

Telegram Trigger: Activates when users send messages to the configured Telegram bot. The trigger captures message content and user information to initiate the workflow.

Nodes Used¶

Node Type	Purpose
Telegram Trigger	Receives incoming messages from users
HTTP Request (PromptLayer)	Fetches the production-tagged prompt template
LangChain Agent	Core AI bot that processes messages and generates responses
OpenRouter Chat Model	Google Gemini 2.5 Flash Lite language model for AI responses
Airtable Tool	Searches project database by term, grade, and subject
Buffer Window Memory	Maintains conversation context with 1-message window
Telegram Send	Delivers bot responses back to users
HTTP Request (Logging)	Logs interactions to PromptLayer for analysis
Webhook	Additional endpoint (purpose unclear from current connections)
Google Gemini Embeddings	Text embeddings (not currently connected)

External Services & Credentials Required¶

PromptLayer¶

Purpose: Collaborative prompt management and interaction logging
Credentials: API key for prompt retrieval and logging
Configuration: Prompt template named "pbapldemo" with "prod" label

OpenRouter¶

Purpose: Access to Google Gemini 2.5 Flash Lite model
Credentials: OpenRouter API credentials
Model: google/gemini-2.5-flash-lite

Telegram¶

Purpose: Bot messaging platform
Credentials: Telegram Bot API token
Setup: Bot must be created via BotFather

Airtable¶

Purpose: Project database storage
Credentials: Airtable API token
Database: "PBA Virtual Bot Project" base with "PBA Projects' Title and Description" table

Google AI¶

Purpose: Text embeddings (if enabled)
Credentials: Google Palm API key

Environment Variables¶

No explicit environment variables are defined in this workflow. All sensitive data is managed through n8n's credential system:

OpenRouter API credentials
PromptLayer API key
Telegram Bot token
Airtable API token
Google AI API key

Data Flow¶

Input¶

User Message: Text message from Telegram user
User Context: Telegram user ID and chat information
Prompt Template: Latest production prompt from PromptLayer

Processing¶

Message Analysis: AI agent processes user intent
Database Query: Searches Airtable for relevant projects based on term/grade/subject
Context Integration: Combines current message with conversation history
Response Generation: AI creates contextual response

Output¶

Bot Response: Formatted message sent back to user
Interaction Log: Complete conversation logged to PromptLayer with metadata

Error Handling¶

Support Bot Node: Configured with retry on failure (2-second wait between attempts)
Memory Management: Uses session-based memory tied to chat ID to prevent cross-user contamination
API Resilience: Multiple external service dependencies may require additional error handling

Known Limitations¶

Memory Window: Only retains 1 previous message, limiting conversation context
Single Model: Relies solely on Google Gemini 2.5 Flash Lite
Database Dependency: Project information accuracy depends on Airtable data quality
Prompt Management: Changes to production prompts immediately affect all users

No related workflows identified in the current context.

Setup Instructions¶

1. Import Workflow¶

Import the workflow JSON into your n8n instance
Ensure all required node types are available

2. Configure Credentials¶

Create the following credentials in n8n:

OpenRouter API - Name: "OpenRouter Org" - Add your OpenRouter API key

Telegram Bot API - Name: "PBA PL Demo" - Create bot via @BotFather on Telegram - Add bot token

Airtable API - Name: "pba" - Add Airtable personal access token - Ensure access to base "appntXVw3V83JQuiW"

Google Palm API - Name: "Waringa Voice" - Add Google AI API key

3. Set Up PromptLayer¶

Create account at PromptLayer
Create prompt template named "pbapldemo"
Tag desired prompt version with "prod" label
Update API key in HTTP Request nodes

4. Configure Airtable Database¶

Ensure your Airtable base contains: - Base ID: appntXVw3V83JQuiW - Table: "PBA Projects' Title and Description" - Required fields: Term, Grade, Subject, project titles and descriptions

5. Test and Activate¶

Test the workflow with sample messages
Verify all external service connections
Activate the workflow to enable the Telegram trigger

6. Webhook Configuration (Optional)¶

If using the webhook node, configure the endpoint path and update any calling systems to use the generated webhook URL.