Hey Product Hunt! I’m Jawad, founder of Pabandi. Today we’re launching a booking ecosystem that fuses machine learning and Web3 to kill the no‑show problem for good — starting with hospitality, salons, clinics, and freelancers. Why this matters: No‑shows bleed over $50 billion a year globally. Businesses lose revenue. Reliable customers get punished with blanket deposits. And a freelancer’s 5‑star reputation on Fiverr means nothing when they book a co‑working desk across town. We saw a broken trust layer — so we built one. What Pabandi does, technically: 1. AI Reliability Score — A stacked ensemble of gradient‑boosted trees, temporal graph neural networks, and a meta‑learner crunches 247+ features (booking history, lead time, traffic, device patterns, and optional on‑chain wallet activity). Accuracy: 91% on held‑out test sets, perfectly calibrated. 2. Web3 Escrow — When a customer books, if the AI flags elevated risk, a refundable deposit is locked in a Solana escrow contract via our Anchor program. The deposit amount fluctuates dynamically with real‑time risk. 3. Guaranteed Revenue — If that high‑risk customer doesn’t show, the escrow automatically releases to the business — income the business would have lost. If they do show, the deposit returns and PAB utility tokens mint directly to their Phantom wallet. 4. Portable Reputation — Users can link LinkedIn, Fiverr, Upwork, and their wallet. Public signals (account age, star ratings, completion rates) are hashed into privacy‑safe features that boost their score — so trust travels across every booking. 5. Staking & API — Businesses stake PAB tokens to access our scoring API at up to 40% off, creating a sustainable token flywheel. The same API can be plugged into any booking platform, insurance underwriter, or marketplace. Early traction: Our pilot in Karachi (salons & clinics) delivered a 67% drop in no‑shows and 34% revenue recovery. Customer satisfaction hit 94%. The waitlist passed 1,000 in a few weeks. Try it: https://pabandi-42c5b.web.app/ I’m here all day to geek out about the ML stack, the Solana contracts, or our plans to scale this across Southeast Asia. Ask me anything! AI bookings. Web3 escrow. No empty seats. Wada karo, inaam pao. 🤝🇵🇰

Maker

📌

2mo ago

How does the reputation score actually work if someone is brand new and has no LinkedIn, Fiverr, or wallet history yet — do they just get flagged as high-risk by default?

The risk-tiered escrow approach is genuinely smart, only penalizing the people who actually cost you money feels way more fair than the usual blanket deposit. Curious how the reputation score handles someone rebuilding after a rough patch.