Team IMP

Inspiration

Inspired by Siri, one of our team members who owns a pre-order bakery.

Small home-based bakeries face an operational complexity and food waste. Independent bakers lose up to 20% of their product to spoilage, cutting roughly 4.8% straight out of their bottom-line margins.

Siri spent hours late at night manually entering crumpled grocery receipts into spreadsheets, struggling to convert bulk purchase weights into per-gram recipe costs. Even worse, unpredictable quiet sales weeks meant fresh butter and berries ended up in the trash bin. We built BakeWise to eliminate spreadsheet friction, automate COGS math, predict demand dips, and transform perishable stock into 1-click sales deals before it expires.

What it does

BakeWise is an automated intelligence engine designed specifically for home-baking kitchen environments. It transforms kitchen operations through a 4-step workflow:

Scan & Verify (OCR Pantry Upload): Siri snaps a photo of her store receipt. Our AI parses line items instantly. Any smudged or ambiguous items are highlighted in a Yellow Verification Drawer for a 5-second edit before saving.
Batch COGS & Margin Pricing: The math engine converts bulk packages into exact per-gram recipe costs using the formula:

$$\text{Unit COGS} = \left( \frac{\text{Bulk Price}}{\text{Bulk Mass}} \right) \times \text{Recipe Mass} + \text{Packaging Price}$$

It then automatically enforces a target markup rule (e.g., ( 2.5\times )), displaying projected profit margins instantly.

Demand Forecasting & Waste Alerts: By analyzing 60 days of sales trends alongside day-of-week heuristics, the system flags upcoming slow periods: "⚠️ Friday pre-orders down 35%. 2kg Unsalted Butter expires in 4 days."
1-Click Sales Activation: Instead of feeling demoralized, Siri taps "Dispatch Offer". The app generates a margin-safe promotional deal and dispatches it immediately to customer channels without manual copy-pasting.

How we built it

We engineered BakeWise as a full-stack, sponsor-powered application:

UI/UX Design: Built in Figma with kitchen-friendly, high-contrast visual components and custom color tokens for receipt verification drawers and COGS cards.
App Building & AI Engine: Built using ProjectAAL as our core development accelerator and AI gateway, scaffolding React/Tailwind CSS components and powering our image parsing and demand reasoning logic.
Automation Pipeline: Powered by n8n Cloud Pro. When a deal is activated, a live HTTP POST webhook receives the payload and formats it into a broadcast message pushed straight to Discord and customer channels.
Deployment: Hosted live on Vercel with a custom domain at https://cupcakeprincess.xyz.

Challenges we ran into

Handling "Dirty Receipt" OCR Failures: Real crumpled grocery receipts often produce cryptic SKUs or smudged numbers. We solved this by designing a low-confidence fallback system that flags items with ( <85\% ) confidence in yellow, keeping data 100% accurate without slowing down the baker.
Bulk-to-Gram Conversion Math: Reconciling inconsistent bulk store units (kg, lbs, dozens) against granular recipe weights (grams, single units) required backend math logic before feeding data into our ( 2\times - 3\times ) markup engine.
Zero-Friction Distribution: Eliminating manual copy-pasting forced us to configure an end-to-end webhook bridge between Vercel and n8n Cloud that fires formatted payload cards in under 1 second.

Accomplishments that we're proud of

Successfully taking an application from initial concept to a live, production-ready deployment at https://cupcakeprincess.xyz.
Building a real, end-to-end automated loop where clicking a button on a web app instantly triggers a live formatted broadcast card via n8n Cloud.
Creating an authentic, zero-friction user experience that solves a real business problem for our domain lead, Siri.

What we learned

How to leverage AI app builders like ProjectAAL alongside workflow automation platforms like n8n to build fully functional prototypes at high speed during a sprint.
The importance of micro-step UX design—building software that works effortlessly in a fast-paced kitchen environment.

What's next for BakeWise

Multi-Channel Dispatch: Expanding our n8n webhooks to include direct WhatsApp, Twilio SMS, and Instagram DM automated marketing.
Real-Time Price Scraping: Integrating Firecrawl to scrape live supermarket grocery prices to alert bakers when ingredient market prices shift.
POS Reconciliation: Connecting directly with Square and Shopify APIs to auto-reconcile real-time sales against pantry stock levels.