The exchange for the debt economy
Veterans from PayPal’s risk and fraud teams have launched a new startup called Pinch AI, focused on using artificial intelligence to fight retail return fraud and improve customer loyalty.
What Pinch AI Does
-
Pinch AI provides fraud prevention and customer loyalty software aimed at retailers, with a particular focus on detecting fraudulent or abusive product returns.
-
Its platform uses AI to assess customer intent, spot abuse patterns, and automatically make return and refund decisions across checkout, return initiation, and warehouse operations.
Founders and Background
-
Pinch AI was founded in 2023 by three former PayPal executives: Arthi Rajan Makhija, Chirag Vaya, and Jayan Tharayil, all of whom worked on risk and fraud systems at PayPal.
-
The company is based in San Jose, California, and is drawing directly on the founders’ experience building large-scale fraud and risk infrastructure in digital payments.
Funding And Investors
-
Pinch AI has raised about $5 million in seed funding to accelerate product development and go-to-market efforts.
-
The round was led by Dynamo Ventures and Infinity Ventures, with participation from Defined Capital and PayPal Ventures, giving the startup strategic ties back to PayPal’s ecosystem.
Product Focus And Results
-
The startup is building protection models, warehouse intelligence systems, and an adaptive “return engine” that uses AI to evaluate risk and reward trustworthy customers.
-
In one cited case, a North American premium apparel brand reportedly saw an 8% drop in returns and a 20% increase in its key customer group after using Pinch AI’s technology.
Why It Matters
-
Retailers are struggling with rising return fraud and policy abuse, especially as e‑commerce expands and liberal return policies get exploited.
-
Pinch AI positions itself as an intelligent operating system for returns, aiming to replace blunt, one-size-fits-all policies with adaptive, real-time decisions that both cut losses and keep good customers happy.
Upcoming Events
