FANA LLM v0.2.9 - Feedback and Sentiment Mechanism for Telegram Groups

---

Objectives:

• Data-Driven Insights: Analyze customer and team interactions and provide valuable insights to help you make informed business decisions.

• Sentiment Monitor: Analyze customer and team for positive or negative sentiment

• Summarise, and suggest action plans, next steps and more

• Expand the use cases and the conversational AI capabilities of the FANA LLM on messaging applications

• Implement a daily feedback mechanism to provide teams with personalized communication

• Ensure the AI will reply only with specific triggers and remove the need for a "/" at the bot command

  // Only respond to messages that start with 'Hey fana'
  if (text.includes('Hey fana')) {

Technology Used:

• Node.js as the primary programming language

• Express.js as the web framework

• Axios for making API requests

• Telegram Bot API for integrating with Telegram messenger

  // Function to make the query
  async function query(data) {
      console.log('Entering query function');
      try {
          const response = await axios.post(API_URL, data, {
              headers: {
                  Authorization: `Bearer ${process.env.API_TOKEN}`,
                  'Content-Type': 'application/json',
              },
          });
          console.log('Query successful');
          return response.data;
      } catch (error) {
          console.error('Error in the query:', error.message);
          throw error;
      }
  }

• Redis for storing and retrieving message history

• Cron job scheduler for scheduling daily feedback

New Features:

• Improved Telegram integration AI capabilities using advanced natural language processing (NLP) techniques

• Daily feedback mechanism that provides users with personalized summaries and action plans

  // Function to send daily feedback
  async function sendDailyFeedback() {
      console.log('Entering sendDailyFeedback function');
  
      try {
          // Send a command to the backend to generate the feedback
          const result = await query({ question: 'generate daily feedback' }); // Use the entire text as the query
          console.log('Query result:', result);
  
          // Extract the text from the JSON response
          const responseText = result.text || 'Could not retrieve response text';
          console.log('Response text:', responseText);
  
          // Log for outgoing message
          console.log(`Response: ${responseText}`);
  
          // Send the response to multiple Telegram chats
          const chatIds = ['-chatIdHere', '-chatIdHere'];
          for (const chatId of chatIds) {
              await axios.post(`https://api.telegram.org/bot${TELEGRAM_BOT_TOKEN}/sendMessage`, {
                  chat_id: chatId,
                  text: responseText,
                  parse_mode: 'Markdown', // Add this parameter to enable Markdown parsing
              });
              console.log(`Sent response to Telegram chat ${chatId}`);
          }
      } catch (error) {
          // Handle errors
          console.error('Error sending daily feedback:', error.message);
      }
  }
  
  
  // Schedule the daily feedback
  cron.schedule('0 12 * * *', sendDailyFeedback);
  console.log('Scheduled daily feedback to run at 12:00 UTC every day');

• Enhanced error handling and logging for improved system reliability

  console.log('Entering query function');
  console.log('Entering handleTelegramMessage function');
  console.log('Entering sendDailyFeedback function');
  console.log('Scheduled daily feedback to run at 12:00 UTC every day');

Bug Fixes:

• Fixed issue with query processing that caused errors in response generation

• Fixed issue with message history storage that caused data loss

• Improved performance of the system by optimizing database queries

Known Issues:

• None

Future Development:

• Integrate with additional messaging platforms (e.g. WhatsApp, Slack)

• Build a RAG system for chat_history leveraging embeddings and cosine similarity

• Enhance the conversational AI capabilities with more advanced NLP techniques

---