Inspiration
Finding the right parts to build something is harder than it should be. When we want to build a simple object like a table or a chair, we often don’t know what components are required or how to search for them on large industrial websites like Grainger. We wanted to make product discovery feel more intuitive by letting users describe what they want to build instead of manually searching through catalogs
What it does
BuildAssist allows users to describe what they want to build in natural language (for example, “I want to build a table”). The system uses AI to break that idea into required components and then recommends relevant products from a Grainger-based dataset. Each recommended item includes a direct link to Grainger, making it easy to explore or purchase the product.
How we built it
We built BuildAssist using Django for the backend and HTML/CSS for the frontend. The AI logic uses the OpenAI API to extract build components and normalize them into industrial search terms. Product data is sourced from a cleaned Grainger dataset, and the system intelligently searches it to find matches. When no exact match exists, the app still generates Grainger search links as a fallback. We also designed a modern, animated UI to make the experience engaging and easy to understand.
Challenges we ran into
One major challenge was matching user-friendly build descriptions with industrial product names. Another challenge was handling cases where the dataset didn’t contain a direct product match. We also had to deal with API key security and GitHub push protection, which required us to refactor our setup to use environment variables correctly.
Accomplishments that we're proud of
We’re proud that we built a full end-to-end AI-powered system—from natural language input to real product recommendations with external links. We also created a polished, interactive UI that feels like a real product rather than just a prototype. Successfully handling security best practices for API keys was another big accomplishment.
What we learned
We learned how to integrate AI meaningfully into a real application, not just as a demo feature. We also gained experience with Django, dataset-driven search, GitHub workflows, and secret management. Most importantly, we learned how to design around real-world constraints, like incomplete data and security requirements.
What's next for Untitled
Next, we want to group products by component, add relevance scoring, and allow users to save or compare builds. We also plan to integrate real-time APIs instead of static datasets and deploy the project so it can be used outside a local environment.