Inspiration

Service work is the largest labor category on Earth — over 100 million people worldwide. Most juggle multiple jobs, memorize dozens of menus, adapt to constantly changing environments, and are expected to deliver premium, personalized service every single shift. Turnover is sky-high, job stability is low, and yet guest expectations keep rising.

We realized something: no one is building tools for the workers themselves. POS systems serve owners. AI scheduling tools serve management. But nothing exists to make workers faster, smarter, more confident, and more financially secure.

So we built SWANS — the Service Work Assistant & Navigation System. Not to replace workers. But to make workers superhuman.

What it does

SWANS is an AI/AR copilot that follows a service worker across every gig, restaurant, lounge, bar, café, or venue. It:

Converts live table conversations into structured orders using AI

Displays this data on a subtle AR HUD for the worker

Tracks regulars and “SWANS” (VIPs) to elevate guest experience

Helps workers avoid mistakes and reduce cognitive load

Positions floating indicators above guests to guide re-engagement (WIP)

Eliminates the need for pen+pad, POS terminals, or memorization

Empowers workers to deliver premium service anywhere

Helps workers increase tips, speed, and consistency

The result: Faster service, fewer mistakes, higher tips, and premium hospitality delivered by anyone, anywhere.

How we built it

Built a Unity-based AR HUD optimized for subtle, non-intrusive overlays

Used live audio input → AI → structured JSON schema for orders (items, modifiers, quantity, category, VIP tag, special instructions)

Displayed the parsed order in real time on the HUD

Added the first version of S.W.A.N. Indicators — small floating markers meant to attach to unique faces

Created a prototype system for tracking regulars and classifying “SWANS” (our VIP tier)

Attempted automatic table detection to place “engagement markers” in AR space

Experimented with face anchoring to position markers above individuals

Everything was optimized around one principle: augment the human, don’t interrupt them.

Challenges we ran into

Automatically detecting tables in complex lighting and seating layouts

Getting floating engagement markers to anchor stably in physical space

Face detection for positioning SWAN indicators above heads without drift

Creating a clean, minimal UI that enhances service without distracting from it

Ensuring structured order output stayed consistent across varied conversation styles

Avoiding latency when updating the AR HUD with new order data

Designing a worker-focused tool that respects privacy and portability

Accomplishments that we're proud of

Built end-to-end “voice → structured order → HUD display” working inside AR

Designed a consistent, extensible schema for order capture

Implemented the first version of VIP/SWAN tracking

Established a visual language for floating face indicators

Made AR interfaces that preserve human connection by staying subtle and elegant

Proved the viability of a worker-first service copilot

Demonstrated a shift from enterprise tools → personal augmentation tools

What we learned

Workers don’t need automation — they need augmentation

Personal AI adoption is faster than enterprise AI adoption

The best AR UI is nearly invisible and preserves eye contact

Structured outputs + AR overlays = reliable, real-time utility

Building a universal “service OS” requires supporting multi-job workflows

The future of frontline labor belongs to tools that travel with the worker, not the employer

What’s next for SWANS?

Expand the worker profile system to follow workers across venues

Improve marker stability with better face anchoring + depth understanding

Build out a name-recognition & regular-guest memory system

Add micro-gestures for confirmations/corrections

Layer in tip-optimization & service-speed analytics

Push toward a worker subscription model (Superhuman for service workers)

Pilot with multi-job workers across restaurants, lounges, bars, and hotels

Move to lightweight AR form factors and eventually to everyday wearables

SWANS is just the beginning. We’re building the world’s first AI exoskeleton for service workers — a universal copilot that democratizes premium service and helps workers earn more, stress less, and perform at a superhuman level.

Built With

Share this project:

Updates