What to Build

Pick one industrial surface. Ship a system a plant, a supplier desk, or a railroad could actually run. Stacks below are starting points, not a mandate.

AI for Manufacturing

Computer vision and robotics on the plant floor. Treat the line the way a grocery store treats self-checkout: overhead cameras watching every station, catching missing parts, defects, unsafe motion, and stalled work before they become scrap.

Where to aim:

• Overhead and station cameras for tracking, spotting, and monitoring parts, tools, and people on the line
• In-process quality: wrong assembly, missing fasteners, surface defects, cycle-time stalls
• Safe robot and drone VSLAM through a live factory — localize, map occupancy, move without colliding, keep a human in the loop
• Generative CAD: propose and iterate parts a shop can actually cut, print, or machine — manufacturable geometry, not render-only form<
• CAD to production and work instructions: turn a released model into travelers, setups, toolpaths, and station-level instructions the floor can follow

Supplier Relationship Management

Make the supplier graph, the order line, and the payment the same living system. Knowledge graphs keep relationships current. An API-first MES/ERP — or a clean adapter — publishes production and order status upstream and downstream. Modern payment rails, including stablecoins where they shorten time-to-pay, close vendor ops.

Where to aim:

• A knowledge graph of suppliers, parts, certifications, and live order state
• API-first MES/ERP, or an adapter, sharing production and order-line expectations with customers and suppliers
• Vendor ops and settlement: approvals, reconciliation, and working-capital velocity — including stablecoin rails where they beat correspondent-bank delay

AI for Transportation

Computer vision for track and railroad maintenance. Inspect at network scale, rank the work, and keep people in the loop.

Where to aim:

• Track geometry, fastener, and surface inspection from locomotive, wayside, or drone cameras
• Defect detection that flags, localizes, and ranks work before it becomes a slow-order or a failure
• Predictive maintenance that turns vision plus sensor history into a crew-ready work order

Open Forum

Bring ideas and solutions that aren't already in the brief. This is the lane for work at the intersection of technology, manufacturing, and AI that the other tracks don't cover — problems nobody is focusing on yet, shipped as a system someone could actually run.

Where to aim:

• An overlooked industrial or civic problem at the intersection of technology, manufacturing, and AI
• A working prototype a plant, a crew, or a city could try — not a restatement of the three named tracks
• A clear operator: who runs this after Demo Day, and why it isn't already standard practice

Every team must submit:

1. Problem track. Name which of the three tracks you submitted toward: AI for Manufacturing, Supplier Relationship Management, or AI for Transportation.

2. GitHub link. Include a public repository for the working solution.

3. Video demo. Include a brief recording of the system running end to end.

4. Talent registration. Confirm every participant is registered in the Fort Worth DAO Talent Database at https://talent.fwtx.city.

5. In-person Demo Day. Every team must attend Demo Day in person for judging: https://luma.com/ygyxb4yy.

Ship it like a product. Mind the ergonomics of your code. Teammates, judges, and a plant engineer should be able to clone, run, and maintain it. Prefer a thin end-to-end path — capture, decide, act — over a pile of notebooks. README, environment, and a demo that boots. Think like a bootstrapped startup taking a pilot to a floor, not a weekend prototype that dies on a laptop.