Sports
Tunnel Solutions
AI match intelligence for the 2026 FIFA World Cup — live from the tunnel. Built by rs9459.
A chat interface backed by a Python agent that combines live web news via Nimble, structured match data from the Football API, and Groq's LLaMA model to deliver sharp, grounded analysis across three modes: pre-match scouting, live tactical reads, and post-match breakdowns.
Quick Start
Inputs
- Match question Any natural-language question about the 2026 World Cup — a matchup, a player, a fantasy call, a live score, or a post-match breakdown.
- Mode Auto-detected from the question, or manually set: Pre-Match, Live, or Post-Match.
Outputs
- Pre-match tunnel report: team form, key battles, injury news, predicted lineup notes, fantasy captain picks
- Live analysis: score, momentum, key events, tactical shifts
- Post-match breakdown: goal analysis, player ratings, fantasy point summary
- Tournament desk answers: power rankings, golden boot leaders, group stage standings
- Sources attribution per response
How it works
The agent operates in three modes, auto-detected from the user's question or manually selected in the UI. Each response includes a sources list showing which data providers contributed — "Nimble Web Intelligence" and/or "Football API" — so the user knows what the answer is grounded in.
- PRE-MATCH Triggered by questions about upcoming matches, lineups, form, or fantasy picks. The agent calls Nimble Web to scrape live Google search results for the team or matchup mentioned, fetches the next 5 World Cup fixtures from the Football API, and passes both to the LLM. Every pre-match response opens with the signature "🚇 TUNNEL REPORT —" format.
- LIVE Triggered by keywords like "right now", "live", or "current score". The agent fetches all currently live World Cup fixtures from the Football API alongside fresh Nimble web results, then generates a real-time tactical read — momentum shifts, key events, substitution signals.
- POST-MATCH Triggered by questions asking for results, breakdowns, or what happened. Combines Nimble web results (match reports, ratings coverage) with the LLM to produce goal-by-goal breakdowns, player ratings, and fantasy point analysis.
Stack
Nimble primitives plus the full runtime stack.
Nimble APIs
What it does
- Nimble Search Scrapes live Google search results for any match query — surfacing news, injury updates, squad announcements, and tactical previews in real time. Feeds the agent's context before every LLM call.
3rd Party Tools
Role
- Python 3.12 Backend agent and API server
- FastAPI + uvicorn REST API (POST /api/chat) serving the frontend
- Groq (
llama-3.3-70b-versatile) LLM — generates the match briefing from live context - API-Sports Football API Structured match data: upcoming fixtures, live scores, player stats for the 2026 World Cup
- Vanilla HTML/CSS/JS Single-file frontend with a football pitch theme, anime tunnel intro animation, sidebar quick prompts, and a three-mode selector
- Railway Backend deployment target (already live at tunnelsolutions-production.up.railway.app)
Reach out if you have any questions.
Talk to an Expert