Vendoract | Devpost

Inspiration

Choosing the right external vendor tools is time-consuming, overwhelming, and often based on biased opinions or outdated reviews. I wanted to build something that empowers tech owners and decision-makers to make informed product choices — quickly, accurately, and with minimal effort — using the power of generative AI.

What it does

Vendoract is an interactive, AI-powered assistant that helps users discover and compare external products (like cloud platforms, databases, CRMs, etc.) based on their specific needs. The user simply selects their preferences through a few dropdowns and checkboxes, and the app generates a detailed, well-researched report — powered by Gemini LLM and live web data — comparing tools across crucial factors like stability, cost, security, scalability, public opinion, and long-term benefits.

How I built it

I built a responsive frontend using Next.js and Tailwind CSS that guides the user through a step-by-step preference selection process. On the backend, I used FastAPI to handle logic and requests, BeautifulSoup for live web scraping of tool reviews, and Gemini (via Google Generative AI API) to generate the final reports. Everything is containerized with Docker, and the entire system runs using Docker Compose.

Challenges I ran into

Making the LLM generate relevant and unbiased output with minimal prompts was tricky — I had to fine-tune the prompt structure.
Ensuring CORS compatibility between frontend and backend services.
Providing a smooth, intuitive user experience with minimal user input and zero friction.

Accomplishments that I'm proud of

Creating a full-stack AI solution that produces accurate, readable, and helpful product reports in under 30 seconds.
Achieving a clean and responsive UI that works across devices.
Fully containerizing the app so it's easy to run anywhere.
Giving users a print-to-PDF option to save or share the reports.