Red Alpha assessment context · Assessment prototype; prepared for review
RouteBrief: a travel check that shows its sources
A local prototype that brings a U.S. air route, travel date, weather and airport notices into one view. I directed the Codex-assisted build; it is prepared for review and an attended demonstration.
01 / The user problem
One place to start a travel check.

A person helping a traveller may need to check several weather and airport websites. RouteBrief brings that first check together around the route and travel date.
- Spot possible disruption.
- See the source behind each concern.
- Know what to check next.
An unpaid assessment prototype, prepared for review. It has not been submitted.
02 / The existing workflow
Choose the trip. See what needs checking.

Choose supported U.S. departure and arrival airports and a travel date. The local app brings together National Weather Service forecasts and alerts with current Federal Aviation Administration (FAA) airport events.
- A short assessment beside the information behind it.
- Clear notice when a source is unavailable or does not cover the requested date.
- Clearly labelled sample scenarios for a demonstration that can be repeated.
The app runs locally. A public live demonstration has not been set up.
03 / Decisions and uncertainty
Know what the information can tell you.
An airport notice tells us about conditions now; a weather forecast concerns a future date. RouteBrief keeps those separate and shows when information is missing or the requested date is beyond the available forecast.
- Use each airport’s local date when checking the weather.
- Keep source links and timestamps beside the assessment.
- Show a current airport event as current context, rather than a prediction for the travel date.
- Clearly label made-up scenarios as demonstration data.
The prototype highlights possible disruption. It cannot predict whether an individual flight will be cancelled.
04 / How AI was used
AI helped me build it. The sources support the check.
I directed the project and used Codex to create the code, documentation and tests. When the app runs, it follows defined rules and checks retrieved evidence; it does not call an AI language model to make the assessment.
- The build record dated 4 November reports 25 automated tests passed.
- It also records checks using live sources, made-up scenarios and dates beyond the forecast range, plus desktop and phone-width inspection.
- My next step is to review the prototype and rehearse the demonstration.
These are recorded checks from the prototype build. My own review remains a next step.
05 / Attended walkthrough
Walk through a trip. See the evidence and limits.
In an attended demonstration, we can check a live route, change the travel date and try a labelled sample scenario, with the source information visible throughout.
- Checks supported U.S. airports for the travel day as a whole.
- Does not predict a specific flight’s cancellation or automatically rebook a trip.
- The existing build record establishes neither public hosting nor results from operational use.
Prepared for review; the assessment assignment has not been submitted. Ask me for an attended demonstration.
Use the arrows or keyboard ← →. Present simplifies the view; Esc returns. Print includes every slide.
Ask me to walk you through this ↗