AI Enterprise SolutionsGPT & Claude
Locked In
The user: Mid to large scale markets and grocery stores with high rush hours. The problem: A high rush hour leads to negative conversion when the person coming to get the groceries are already tired. This leads to loss of revenue if considered precisely.
Preliminary score
55.0/ 100Promisingaverage of 2 judges
- Value for the user17.0 / 25
A specific problem and a clear outcome from solving it.
- Prototype and use of AI13.5 / 30
A working core scenario, and what AI actually contributes to the result.
- Quality testing12.5 / 20
Tests, examples of failures, and comparison with the current approach.
- Feasibility7.5 / 15
Data requirements, running costs and a clear next step.
- Originality4.5 / 10
A meaningful difference from common solutions.
What the judges wrote
- GPT-6 AstraPreliminary60/100
- Strengths
- Your single-QR checkout targets a concrete grocery-store bottleneck and provides a clear checkout-time metric. Your README documents a reproducible basket-to-checkout workflow, server-side totals, and transparent hardware simulation. Your testing notes include specific camera failure conditions, end-to-end test coverage, and an A* comparison against BFS.
- To improve
- Your AI contribution needs a named forecasting method and held-out results against a simple baseline; synthetic analytics alone do not establish its decision-making value. Reconcile the reported barcode database failure with the README's working-flow description, and repeat the checkout timing comparison across multiple baskets, including removals and missed scans. Your next step should be a costed store pilot covering scanner/display hardware, basket-location sensing, POS integration, and ongoing operations.
- Claude Opus 5.5Preliminary50/100
- Strengths
- Your README shows a coherent, working flow: items scanned in the basket appear in a live total, checkout reads a single basket QR code, and the basket resets after payment. That flow is backed by an API, a deployed demo and a phone-camera scanner fallback. Your 60 pytest tests cover EAN-13 check digits, the full purchase flow, store-map integrity and A* versus BFS shortest paths, which is real engineering testing. Your camera measurements (the barcode must fill 50–80% of the frame; it fails at 30% on 640p) and the 47 s versus ~8 s checkout comparison are the kind of concrete evidence judges look for.
- To improve
- Your deck says AI is not the core, and the forecasting and ML page runs only on synthetic data, so AI currently looks secondary. Show one AI feature, such as demand forecasting or stock-risk alerts, working on realistic data, and measure how accurate it is. Explain what makes this different from existing smart carts like Amazon Dash Cart or Caper, and address hardware cost per basket, theft and mis-scans, since in-basket scanners are the hardest part to deploy. Align your submission text with your repo: the write-up lists Next.js and Supabase while the README describes Vite, FastAPI and Neon, and the written testing section undersells the work you actually did.
The team
Sacha Kuenga
designer
Nihad GasimovLead
developer
Ziya Abdullayev
data