GuideMe: Travel Planning — Alexa+ Simulation

Inspiration

Planning a trip often means switching between itineraries, weather forecasts, currency converters and scattered notes. I wanted travelers to refine a plan through a conversation while keeping their earlier preferences.

GuideMe existed before this hackathon. In October 2026, I added a separate, working browser-based Alexa+ experience simulation.

What it does

Travelers can describe a trip in English or Arabic, ask follow-up questions and revise an itinerary. The assistant can call live weather and currency tools and include their sources and dates.

Conversations are saved in the same browser and can be downloaded as text notes. Optional browser speech playback reads responses aloud.

This prototype simulates an Alexa+ travel-planning experience. It is not connected to Alexa devices and does not implement an Alexa SDK integration or an MCP server. It does not make bookings or purchases.

How I built it

The new interface uses HTML, CSS and JavaScript. A Node.js Netlify Function connects to Anthropic’s Messages API and handles model-selected tool calls. Open-Meteo supplies weather forecasts and location search; Frankfurter supplies reference exchange rates.

API credentials remain server-side. Private reviewer access allows testing without a paid subscription.

ChatGPT (Codex) assisted with implementation, debugging, tests, documentation and submission preparation.

Challenges

An early response retrieved weather without completing the requested itinerary revision. I refined the instructions and retested the conversation.

Generated budgets, venue information and accessibility claims still need independent verification. Weather forecasts have a limited time horizon, and exchange rates are reference rates rather than transaction quotes.

Testing and accomplishments

Live browser tests covered trip planning, a follow-up itinerary revision, currency conversion, restoring a six-message conversation after reload, downloading notes and mobile layout.

Additional automated tests covered access control, error handling and safe text rendering. These checks verify specific scenarios, not the accuracy of every generated answer.

What I learned

Testing complete conversations revealed issues that isolated API checks missed. Clear source attribution, preserved conversation context and honest limits are essential for a useful travel assistant.

What is next

Improve budget validation, verify venue and accessibility information, and explore cross-device continuity. A native Alexa integration would be a separate development step.

What's next for GuideMe: Travel Planning — Alexa+ Simulation

Running and judging instructions: README

Built With

• anthropic
• css3
• frankfurter
• html5
• javascript
• netlify
• node.js
• open-meteo