How can a business personalize a digital experience without collecting unnecessary customer data, and who is responsible when an AI agent acts on a shopper’s behalf?

In this episode of Business Tech Perspectives, I speak with Sapphire Samiullah, Product Lead at Pion, about why digital identity is moving from a specialist check at registration or checkout into the infrastructure behind customer trust, loyalty and fraud prevention. Pion supports verified offers across 10 consumer groups in over 100 countries, including students, graduates and workers in eligible professions.

Sapphire explains how a reusable verification status can reduce repeated data collection. A customer who has already proved eligibility can enter a phone number on a participating brand’s site, allowing Pion to return a simple verified status. The brand receives the result it needs, while the customer avoids sharing the same supporting information again. She also describes why adding SMS-based multi-factor authentication made students feel safer rather than creating the expected friction.

We discuss the measures that connect identity systems to business results. Registration and verification conversion rates reveal whether people can complete the process. Method, channel and market comparisons expose weak points. Pion also tracks activation, defined here as a verified user issuing a first discount code within 24 hours, to connect successful verification with customer action.

The conversation then turns to agentic shopping. Existing fraud controls often assume a human is behind each request. An AI agent introduces another test: is it authorized to make this purchase for this person at this moment? Sapphire discusses the need for standards around delegated authority, consent and agent credentials as automated purchasing becomes more common.

Privacy and inclusion remain central. Pion asks for the minimum information required, tells users why it needs each field and encourages people to redact irrelevant details from submitted documents. Multiple verification routes, including human review, can also prevent automated systems from excluding apprentices or others without standard records.

Can companies build identity systems that recognize both human customers and authorized AI agents while preserving consent, inclusion and trust?