Inspiration
We were inspired by how hard it can be to pick the right laptop with so many options and confusing specs. We wanted to make it easier using AI—basically, a smart bot that understands what you need and recommends the best laptops without all the tech jargon.
How we built it
We started by scraping laptop listings and specifications from the Digitec website to form a comprehensive dataset. We then created a filter system to categorize and rank laptops based on factors like performance, price,specs, user profile,gpu, portability, etc... Multiple layers of LLMs were fine-tuned to interpret user prompts and map them to this filter system. Finally, we integrated the system into a Telegram bot that responds to user queries with tailored laptop recommendations.
Challenges we ran into
Parsing and cleaning scraped data to ensure consistency and accuracy. Filtering and feeding the filtered data to our bot. Integrating everything into a user-friendly Telegram interface with real-time responses.
Accomplishments that we're proud of
Successfully scraping and structuring a real-world e-commerce dataset. Fine-tuning a language model that can behave like a knowledgeable tech consultant. filtering the data correctly, based on the current market of PC components. Building a fully functional Telegram bot that bridges natural language and product search.
What we learned
Effective ways to combine web scraping, NLP, and chatbot development. The importance of data quality and user experience in AI-powered tools.
##possible openings We put some functionalities that would allow to extend this project to laptop recommendation to any type of product, given that a dataset is provided (more genericity)
Log in or sign up for Devpost to join the conversation.