Arival AI Lab · Valencia 2026
Turn reviews into business intelligence.
Seed data and prompts for the Arival AI Lab on structuring tour reviews into a reusable intelligence layer for product, marketing, guide development, and competitor analysis.
The idea
Your reviews are probably the richest dataset in your business that you are barely using.
Asking AI to “analyse my reviews” is too vague. Decide what you want to learn, design the schema, then let AI structure the mess.
Downloads
Session files
Start with the lab pack. Use the combined CSV plus the prompt sheet during the session.
What changes
When every review gets analysed, the useful signals stop hiding in the pile.
This lab is a compressed version of the ReviewMine idea: reviews are a living dataset for sales, marketing, product, guide development, and competitive positioning.
Objection mapping
Find the anxieties guests mention before or after booking: safety, pace, value, children, crowds, weather, or confidence.
Testimonial extraction
Pull the proof that sells: the exact guest language that answers a buyer's hesitation or aspiration.
Marketing copy
Turn repeated guest language into landing page sections, advisor emails, ad angles, FAQs, and proposal copy.
Pattern detection
Spot recurring product themes, service issues, seasonal shifts, and guide moments that would disappear in manual review reading.
Guide performance signals
Track named praise and review velocity carefully, using it as evidence of memorable guest outcomes rather than a blunt score.
Competitor benchmarking
Analyse competitor reviews through the same schema, then compare positioning, complaints, emotional outcomes, and underserved guests.
Live session
How to use the files
All data is synthetic and anonymised. It is designed to feel believable without using real customer reviews.
1. Download the lab pack
Use the ZIP if you want everything in one place. It includes the combined CSV, individual CSVs, prompt sheet, and README.
2. Upload the combined CSV
Open Claude, ChatGPT, Gemini, or another file-upload LLM. Upload `arival_review_intelligence_demo_data.csv`.
3. Paste the dashboard prompt
Open the prompt sheet and use the main Review Intelligence Dashboard prompt first.
4. Choose one follow-up path
Use guide velocity, competitor gap analysis, or psychographic proof retrieval depending on what matters most to your business.
Example output
This is what the data can become.
In the lab, the CSV is only the starting point. The useful output is a review intelligence dashboard: themes, objections, proof quotes, guide signals, and competitor gaps that an operator can act on.
Rows analysed
342
operator, competitor, and guide activity rows
Top signal
Personalisation
guests repeatedly value thoughtful pace and tailored delivery
Guide signal
Velocity
named praise per tour, treated as a coaching signal
Competitor gap
Less personal
competitors often win on convenience but lose emotional depth
Objection mapping
Cycling safety, pace, children staying engaged, crowds, choosing the right tour
Proof retrieval
Pull the review quote that matches the buyer's hesitation, aspiration, or decision moment
Guide development
Find guides who generate named, memorable praise and inspect why
Competitor benchmarking
Compare operator and competitor reviews through the same schema
Schema
The schema is where the intelligence lives.
CSV is the spreadsheet version. JSON is the structured object version. The file format matters less than the field design.
ReviewMine
Based on work from ReviewMine.
ReviewMine is my review intelligence project for tour and activity operators. The lab borrows from that work: guide review velocity, competitor gap analysis, and using structured review data to surface the right proof for the right buyer.