snap2wear

"An AI-native bespoke fashion operating system combining precision photo measurement, autonomous tailor dispatch, and zero-waste custom apparel manufacturing."

Comment

Inspiration

The global fashion industry produces over 92 million tons of textile waste each year. The primary driver? Sizing failure. Standardized "Small, Medium, Large" sizing leads to a staggering 30%+ e-commerce return rate, with billions of dollars worth of garments dumped straight into landfills. Meanwhile, traditional bespoke tailoring—the gold standard of zero-waste, perfect-fitting clothing—remains an exclusive luxury confined to elite districts like London's Savile Row.

We built Snap2Wear to bridge this gap: an AI-native operating system that brings bespoke tailoring to everyone. By combining cutting-edge computer vision, autonomous Gemini multi-agent workflows, master tailoring education, and gasless Web3 payments, Snap2Wear transforms custom apparel into an accessible, sustainable, and scalable global standard.

What it does

Snap2Wear is a complete bespoke fashion technology ecosystem:

  1. SnapMeasure AI (Computer Vision Body Sizing):

    • Users upload just two photos (front and side).
    • Our vision pipeline (Google MediaPipe + OpenCV) extracts 33 skeletal landmarks and calculates 9 precision anatomical tailoring measurements in seconds.
    • Agent 2 (Gemini Vision Auditor) inspects photos for poor lighting, heavy shadows, and baggy clothing, giving instant corrective feedback.

  2. 6-Agent Autonomous AI Operating System:

    • Agent 1 (Customer Support & Action Engine): Uses Gemini Function Calling to autonomously query real-time Firestore order tracking, trigger password resets, and check student progress.
    • Agent 2 (Photo Quality Auditor): Evaluates image suitability before committing body dimensions to production.
    • Agent 3 (Savile Row Craft Vision Grader): Inspects student tailoring practice photos across 4 artisan criteria (seam alignment, stitch tension, pressing flatness, neatness), grading from 0–100% and unlocking subsequent modules upon scoring 70%+.
    • Agent 4 (Smart Proximity Dispatch Concierge): Autonomously routes orders to the nearest available artisan tailor, evaluating garment complexity, tailor ratings, and current workload to set dynamic acceptance countdown windows (5–20 min).
    • Agent 5 (Cart Recovery Sales Agent): Analyzes student LMS history to send personalized, encouraging craftsperson recovery messages.
    • Agent 6 & Unified Multi-Task Orchestrator: Conversational order builder that analyzes photos, triggers security actions, updates database records, and generates live payment checkouts in a single multi-step pass.

  3. Savile Row Academy LMS:

    • A 4-week, 23-lesson interactive masterclass taking students from foundational stitches to hand-crafting a complete three-piece suit.
    • Includes an embedded Gemma 4 model running locally in the browser so students in low-connectivity regions can learn with an AI tutor 100% offline.

  4. Multi-Rail Checkout (Fiat & Gasless Web3):

    • Seamless local card and bank payments powered by Paystack.
    • 100% gas-sponsored Circle USDC payments on Polygon Mainnet using Smart Contract Accounts (SCA).

  5. B2B SnapMeasure Developer API Portal:

    • Allows third-party fashion brands and e-commerce stores to integrate our AI sizing engine with unmetered £10.00/month SaaS subscriptions and credit packs via Stripe.

  6. Partner Atelier Marketplace:

    • Allows independent fashion entrepreneurs to launch their own branded virtual bespoke storefronts with custom 10%–25% commission markups.

How we built it

  • AI & Computer Vision: Google Gemini 3.1 Flash-Lite (via Google AI Studio) for multi-agent reasoning, vision grading, and function calling; Google MediaPipe and OpenCV for 3D skeletal landmark extraction; on-device Gemma 4 Web runtime for offline tutoring.
  • Backend & Cloud Architecture: Python 3.12, Flask, Gunicorn, deployed serverless on Google Cloud Platform (Google Cloud Firestore NoSQL database and Google Cloud Storage for secure V4 Signed URL asset delivery) and Azure App Service.
  • Payments & Web3: Circle Developer-Controlled Programmable Wallets (Polygon Mainnet SCA), Paystack API with dynamic multi-currency conversion, and Stripe SDK.
  • Certifications: ReportLab automated verification engine for instant hackathon and course certificates.
  • Frontend: Modern responsive UI with Tailwind CSS, Cormorant Garamond luxury typography, custom Markdown engine, and Web Audio API synthesizer for tailor alerts.

Challenges we ran into

  • Single-Camera 3D Perspective Calibration: Accurately deriving three-dimensional body circumferences (chest, waist, hip) from two-dimensional smartphone photos required building custom geometric ratio scaling algorithms calibrated against user height.
  • Multi-Task Agent Orchestration: Preventing LLMs from entering repetitive conversational loops required architecting a strict multi-pass execution pipeline that executes all background tasks (photo analysis, password resets, database records, order compilation) programmatically before synthesizing a single, structured concierge report.
  • Gasless Web3 Onboarding: Removing the crypto barrier so everyday shoppers can pay with USDC on Polygon without needing MATIC/POL gas tokens or seed phrases required deep integration with Circle’s gas-sponsored Smart Contract Account (SCA) architecture.

Accomplishments that we're proud of

  • 645 Registered Users onboarded organically during the hackathon (+8,442% growth).
  • 146 Active Paying SaaS Subscribers generating £1,460.00 ($1,898+ USD) in Monthly Recurring Revenue (MRR) at an 88% gross margin.
  • 127 Developer API Keys deployed across third-party applications with 12,086+ API calls/credits.
  • 823 Official Certificates Issued and verified.
  • 48 Verified Artisan Tailors connected to our smart dispatch network.
  • Zero Landfill Waste: Built a working prototype of a fully circular, on-demand bespoke apparel pipeline.

What we learned

  • How to combine deterministic computer vision (MediaPipe) with probabilistic multimodal LLMs (Gemini Vision) to achieve millimeter-grade measurement reliability.
  • How to design autonomous agent systems with function calling that take real-world actions (email, database, blockchain) safely and efficiently.
  • How to leverage Circle's Programmable Wallets to bring the speed and security of stablecoins to standard Web2 e-commerce.

What's next for snap2wear

  • Shopify & WooCommerce Plugins: Launching 1-click plugins for online fashion merchants to embed SnapMeasure AI directly into their product pages.
  • Post-Hackathon Marketing Scale: Deploying our $300 Google Ads grant credit alongside organic LinkedIn company page marketing (linkedin.com/company/snap2wear) to expand our B2B developer API user base.
  • 3D Garment Simulation: Integrating Gemini-powered generative diffusion models to let users visualize custom fabrics and cuts on their exact body model in real-time 3D before a single thread is cut.
  • Expanding Artisan Networks: Onboarding over 500 independent tailoring ateliers across the UK, Europe, and Africa to build the world’s largest decentralized bespoke production network.
Share this project:

Updates

Submission history