AI Enterprise SolutionsGPT & Claude

NeuraX

TourCheck is for tourists and locals who want to plan a day out or a trip in Azerbaijan, in Baku and across the regions. The interface supports Azerbaijani, English and Russian. The problem is that travel information is scattered and hard to trust. Prices, opening hours and addresses live on different sites, and it is rarely clear whether a place will be open at the time someone plans to visit. People also struggle to estimate a budget before they go, and to turn a list of places into a sensible route. TourCheck puts this in one place. It finds real venues nearby, looks up prices and opening hours on the web, and checks whether each place is open at the planned time. When no price can be found, it shows an estimate and clearly marks it as an estimate, instead of presenting a guess as fact. It then orders the stops into a route, shows the distance and travel time, and links to a map. Visitors can also reserve a spot on a tour, and the owner sees these reservations in the admin panel. This is my reading of the code, not an official product description. If TourCheck is built for a specific audience, such as a travel agency, the first paragraph should be adjusted.

Open the demo Video GitHub

Preliminary score

44.0/ 100Earlyaverage of 2 judges

Value for the user15.0 / 25
A specific problem and a clear outcome from solving it.
Prototype and use of AI13.0 / 30
A working core scenario, and what AI actually contributes to the result.
Quality testing5.0 / 20
Tests, examples of failures, and comparison with the current approach.
Feasibility6.0 / 15
Data requirements, running costs and a clear next step.
Originality5.0 / 10
A meaningful difference from common solutions.

What the judges wrote

Claude Opus 5.5Preliminary40/100
Strengths
Your core idea, checking a whole itinerary against opening days and hours and flagging risks such as 'the Carpet Museum is closed on Mondays', targets a real and recurring travel-planning pain. Your setup notes describe a concrete stack: a FastAPI backend with /api/places, /api/enrich and /api/route, web search plus Gemini enrichment, Firebase reservations and an admin panel. Marking prices as estimates when no source is found, instead of presenting guesses as facts, is a thoughtful trust-building choice. Supporting Azerbaijani, English and Russian fits the local audience well.
To improve
Your deck is almost entirely concept boxes with illustrative examples, so add screenshots or a walkthrough of a real itinerary being verified, with sources and timestamps shown. Your only testing note is a list of missing HTML element IDs, which is a code observation rather than a test. Run 20–30 real venues or itineraries, record how often the hours and prices are correct, show the failures, and compare the result with manual Google Maps research. For the Enterprise track, name a business user such as a travel agency or tour operator, a metric like itineraries checked per hour or errors avoided, and the API and search cost per verification. Also make your submission consistent: it lists Claude as the model while the setup requires a Gemini key, and your README contains only a title.
GPT-6 AstraPreliminary48/100
Strengths
Your focus on checking planned visit times against venue hours addresses a concrete travel-planning problem in Azerbaijan. Clearly labelled price estimates and source timestamps are sensible trust features. Your detailed setup instructions describe a plausible implementation, while the report of 16 potentially missing DOM elements provides an actionable engineering finding.
To improve
Your materials would be stronger with a captured end-to-end run showing venue lookup, AI enrichment, a closure warning and a usable route; clarify the Claude listing versus the Gemini runtime. Reproduce the DOM issues in a browser and record expected versus actual results for closed days, conflicting hours, unknown prices and API failures. Define a pilot audience, such as travel-agency staff, and compare planning time and missed closures against manual research. Name your live data providers, estimate per-itinerary cost and latency, and specify a bounded next step such as verified coverage of one district.

The team

Ayxan Qənbərov
developer
Amir Abbasov
developer
Jalə QədirliLead
product

Disclosure

Required by the rules. Judges verify it against the repository.

Models: Claude
Data: json
Components: Firebase
Built during the event: Confirmed

Quality testing

The JavaScript looks up 16 element IDs that do not exist in the HTML. Examples are tourList, fSearch, searchForm, resetBtn and regionPage. The original single file has the same gap, so the split did not cause it. Some of these elements may be created later by the scripts, and some lookups are guarded against missing elements. If the tour search and filter section looks broken or empty on the live page, this is the first place to look.

Worth 20 of the 100 points.

Running it

Python 3.10 or newer A Gemini API key from Google AI Studio Optional for deployment: a GitHub account and a Vercel account Run locally Unzip tourcheck.zip and open a terminal in the tourcheck folder. Create a virtual environment and install the dependencies: python -m venv .venv source .venv/bin/activate # Windows: .venv\Scripts\activate pip install -r requirements.txt Create a file named .env next to app.py (copy .env.example) and put your key in it: GEMINI_API_KEY=your_key_here Start the server: uvicorn app:app --port 8000 Open these addresses in a browser: Site: http://127.0.0.1:8000 Admin panel: http://127.0.0.1:8000/admin.html If the server exits with “GEMINI_API_KEY is missing”, the .env file is missing or in the wrong folder. Deploy to Vercel Push the tourcheck folder to a new GitHub repository. Do not commit .env; it is already in .gitignore. On vercel.com choose Add New → Project and import the repository. Leave the build settings as detected (FastAPI). Under Environment Variables add GEMINI_API_KEY with your key. Click Deploy. Your site will be at https://<project>.vercel.app and the admin panel at /admin.html. After changing an environment variable, redeploy so the change takes effect. Firebase The site stores reservations and region data in Firebase, and the admin panel reads them. If the admin panel does not work on your deployed domain, add <project>.vercel.app under Authentication → Settings → Authorized domains in the Firebase console, and check your Firestore security rules. Project layout app.py FastAPI server and API routes (/api/places, /api/locate, /api/enrich, /api/route) main.py Search, geocoding, routing and Gemini logic used by app.py requirements.txt Python dependencies public/ index.html Main page admin.html Admin panel css/ styles.css, overrides.css js/ translations, regions-data, helpers, storage, app, backend-bridge, region-page, firebase Troubleshooting Blank page or missing styles: check that public/css and public/js were uploaded with the project. Search or enrichment fails on Vercel: open the project’s Deployments → Logs. The usual causes are a missing GEMINI_API_KEY or a function timeout. Slow responses: /api/enrich calls web search and Gemini, so it can take several seconds.

Project ownership

The team keeps the rights to this project. Judges were given access for verification only.