Hospitality / Event Catering

Order Intelligence

the challenge

Event Espresso was guessing cups, milk, staff, and specialty drinks from memory. Over-ordering wasted inventory. Under-ordering killed the event. The knowledge lived in people’s heads, not in the order.

Our Solution

We built Order Intelligence as its own engine inside the catering stack: an ops quiz, a deterministic calculator, and an AI review-and-apply loop.

Ops answers canonical questions on beverages, supplies, staffing, and event type — cups per attendee, oat-milk share, wedding vs corporate multipliers, waste, and buffer. Those answers become order line quantities. Missing or bad answers fall back to the old rules.

OpenAI then suggests adjustments from event context and business knowledge. Managers view, apply, or revert. Nothing writes blindly. A natural-language Q&A box sits on top of the current order.

The process
1
Discovery
1-week intensive requirements gathering and compliance review
2
Design
Wireframes and interactive prototype in Figma within 10 days
3
Development
4 agile sprints with weekly demos and feedback loops
4
Launch
Deployed to AWS with full CI/CD pipeline and monitoring
Technology Used
Django / DRF, OpenAI, Next.js 16, React 19, TypeScript, TanStack Query
The results
  • Live inside BrewTrack at app.wecatercoffee.com for real event orders.
  • Model output is trustworthy because it is reversible and grounded in ops-authored answers.
  • Separated from the ops platform so the AI engine can be judged on its own: quiz → calc → review-and-apply.
The hard part was making model output trustworthy and reversible — view a suggestion, apply it, or revert it.
Unseen Engine
AI / Full-stack Partner

Start a Similar Project

Let's discuss how we can deliver similar results for your business.