Week #3 #

Prototype Features #

This week, we made significant progress in developing the SayNoMore prototype. Our primary user interface is through a Telegram Bot, designed to assist users with their travel planning. Here’s what we have accomplished:

1. Telegram Bot Interface:

   • Proof of Concept: We have developed a basic Telegram bot that can handle user messages and perform core functions essential for travel planning.
   • Basic Functionality: The bot can process user requests, ask for additional information if necessary, and return several travel routes that fit within the user’s specified budget.

2. RequestAnalyzer Module:

   • User Input Processing: The prototype can parse and understand user inputs, extracting key travel details such as destination, departure city, dates, and budget.
   • Multi-step Interaction Handling: The system interacts with users iteratively, prompting for more information when initial inputs are incomplete or ambiguous.
   • Validation and Verification: This module ensures the extracted details are valid and accurate, such as checking for valid cities and realistic dates.

3. API Collector Module:

   • Flight Data Retrieval: The bot can fetch the cheapest and most relevant flight options using the Aviasales API.
   • Hotel Data Retrieval: It also searches for hotel availability and pricing based on user-provided criteria.
   • Data Formatting: Retrieved data is standardized and formatted for easy presentation to the user, ensuring consistency and clarity.

4. Integration of All Modules:

   • We have successfully integrated the RequestAnalyzer, API Collector, and Telegram Bot Interface into a cohesive system.
   • The bot now processes user requests, retrieves relevant travel data, and provides recommendations based on the user’s budget and preferences.

User Interface #

The current UI is a Telegram Bot that facilitates user interaction for travel planning. Key screens and interactions include:

1. Welcome Screen:
   • The bot greets users with an introduction and overview of its capabilities.
2. User Input Collection:
   • The bot prompts users to provide travel details such as destination, departure city, and travel dates.
3. Dynamic Querying:
   • If initial information is incomplete, the bot interacts dynamically to gather additional necessary details.
4. Travel Recommendations:
   • After processing the request, the bot provides several travel routes that fit the user’s budget and preferences, including flights and hotel options.

These interactions ensure that users can plan their travel efficiently and receive tailored recommendations.

Challenges and Solutions #

1. Local LLM is GPU Expensive and Hard to Deploy:

   • Challenge: Running the local LLaMA model requires significant GPU resources and presents challenges in deployment, making it difficult to scale and manage.
   • Solution: We are considering alternatives such as utilizing cloud-based LLM services or REST APIs for LLM tasks. This will reduce the burden on local resources and simplify deployment, ensuring scalability and easier management.

2. API Data is Limited by Aviasales Cache:

   • Challenge: The current API data from Aviasales is limited to their cache, which may not cover all possible user requests in real-time.
   • Solution: We plan to develop a functional prototype demonstrating our bot’s capabilities. With this prototype, we will approach Aviasales directly to negotiate access to their full API, which will allow us to handle a broader range of user requests with up-to-date and comprehensive data.

Next Steps #

As we move forward, our focus will be on enhancing the bot’s capabilities and refining its functionalities. Here’s what we plan to work on next:

1. Feature Enhancements:

   • Advanced User Interactions: Improve the bot’s ability to handle more complex user queries and provide richer responses.
   • Expanded Data Retrieval: Integrate additional APIs to offer a wider range of travel options, including alternative accommodations and local attractions.
   • Budget Management: Enhance budget handling capabilities to provide more tailored and cost-effective travel recommendations.

2. User Experience Improvements:

   • Interactive Feedback: Implement more interactive and user-friendly feedback mechanisms to keep users engaged and informed.
   • Error Handling: Improve the bot’s ability to gracefully handle errors and provide helpful suggestions when issues arise.

3. Interface Development:

   • Web and Mobile Interfaces: Begin the development of web and mobile platforms to provide users with more comprehensive and visually rich interfaces for travel planning.

4. Testing and Refinement:

   • Comprehensive Testing: Continue with rigorous testing of all features to ensure stability and reliability.
   • User Feedback Integration: Collect and integrate user feedback to refine and enhance the bot’s functionalities.

5. Documentation and Knowledge Sharing:

   • Detailed Documentation: Update and expand the documentation to reflect new features and provide clear guidance for users and developers.
   • Team Knowledge Sharing: Foster ongoing knowledge sharing within the team to ensure everyone is aligned and up-to-date on the latest developments.