StockPal

Inspiration

What it does

How we built it

Challenges we ran into

Innovation

Shopping often requires users to switch between recipe websites, pantry lists, supermarket apps, price comparisons, and payment pages. StockPal was inspired by the idea of turning this fragmented process into one trusted conversation while also helping merchants anticipate demand and manage products more effectively.

What it does

StockPal is an AI-powered conversational commerce platform for consumers and merchants. Consumers can describe a meal or household need using text or voice. StockPal creates personalised recommendations based on household size, pantry stock, budget, allergies, dietary restrictions, nutrition targets, and past purchases. It can generate complete recipe ingredient lists, compare similar products, estimate remaining pantry stock, and prepare an editable cart using verified merchants. Merchants can manage catalogue items, prices, images, and stock levels. They also receive automated demand alerts and AI-assisted forecasts to help prepare inventory for the following week.

How we built it

We built StockPal using React, TypeScript, Node.js, and Vite, with Motion for interface transitions and responsive interactions. Supabase provides authentication and user-owned storage protected by Row Level Security. The OpenAI Responses API powers catalogue-grounded conversations, structured recipe generation, product recommendations, and demand forecasting. OpenAI transcription converts recorded voice prompts into text. Agent recommendations are validated against a canonical product catalogue before being shown or added to the cart. The checkout prototype includes cart versioning, explicit approval, delivery-detail validation, biometric verification through platform authentication, and simulated payment processing.

Challenges we ran into

One major challenge was preventing the AI from inventing products, prices, or ingredients. We addressed this with structured responses and server-side validation against verified catalogue data. Voice transcription also required careful handling of recording duration, language consistency, noisy audio, and incomplete transcripts. Another challenge was generating complete recipes while distinguishing ingredients already in the pantry, products available for purchase, allergy conflicts, and unavailable catalogue items. We also had to ensure that cart changes invalidated previous approval without interrupting the overall checkout experience.

Accomplishments that we're proud of

We created an end-to-end conversational shopping workflow that connects household planning, product discovery, comparison, cart management, verification, and checkout. We are especially proud of the dynamic recipe-to-cart experience, adjustable pantry stock estimates, allergy-aware recommendations, voice input, product comparison, loyalty programme, merchant catalogue management, and AI-assisted demand forecasting. We also designed StockPal to be transparent: users can review exactly what they are buying, from which merchant, at what price, before approving any transaction.

What we learned

We learned that effective agentic commerce requires more than a capable AI model. It needs reliable catalogue data, structured outputs, strict validation, explicit consent, and clear boundaries around what the agent can do. We also learned that household context significantly improves recommendations. Pantry inventory, serving sizes, nutrition goals, and allergies can turn a generic shopping assistant into a genuinely useful planning tool. For merchants, conversational commerce becomes more valuable when paired with inventory visibility and actionable demand insights.

What's next for StockPal

The next step is to improve connection of StockPal to live merchant catalogues, inventory systems, delivery providers, and payment platforms. Future improvements include automatic pantry updates, more accurate consumption forecasting, live delivery coordination, multilingual voice support, merchant promotion tools, and privacy-preserving neighbourhood demand intelligence.we're proud of