Perplexity tests Visa integration controls for contextual shopping AI

Right now the clearest signal from Perplexity is a pivot toward personalized commerce. Recent builds of the app reveal a concealed settings panel labeled Edit Third-Party Personalization that includes a toggle for Visa and places user control at the center of the experience.

Behind the settings work sits an upgrade called Super Memory, described as a way to help users retrieve contextual information even though official details remain sparse. In the new panel, Perplexity will only sync personalization tokens from external providers when a user explicitly enables them, and turning a provider off removes its data from the system, which implies the company does not expose a user’s own Perplexity information back to those services. The feature is still hidden and not broadly available, and the presence of the Visa logo alongside references to multiple providers suggests preparation to tap commercial signals such as purchasing behavior. That approach frames consented data as the foundation for personalized commerce.

Visa’s public materials describe an ambition to power secure, AI driven commerce by enabling agents that tailor services using preferences and transaction history, and earlier this year the company discussed how generative AI will be integral to building hyper personalized commerce. If Perplexity connects to such systems, the assistant could surface recommendations or automated insights based on spending patterns inside a tightly scoped and user controlled environment. The net effect would be to anchor personalized commerce in a secure data sharing loop that users can pause at any time.

Across the market, rivals are exploring memory and long term personalization, while Perplexity is signaling differentiation through trusted partnerships and fine grained controls that live in the settings menu where individuals can manage each provider separately. The design speaks to power users as well as professionals in commerce, research, and fintech who want smarter answers without ceding privacy. In competitive terms, the ability to ingest consented third party signals without reciprocally handing over chat history positions the assistant as a pragmatic gateway to personalized commerce.

Global capital has flowed toward platforms that can translate data access into tangible utility, and arrangements with large scale payment networks are often seen as clearance to operate at consumer scale. If this hidden integration ultimately ships, it would align with the broader investment narrative that ties privacy controls to adoption, a theme likely to resonate from Silicon Valley product teams to European regulators. For investors and founders alike, the lesson is that personalized commerce advances fastest when consent, portability, and revocation are built in from the start.

The practical consequence for enterprises is a shift from generic chatbots to context aware agents that can draw on permissioned tokens to generate training materials, support replies, and purchasing suggestions with far less manual setup. By binding memory to vetted transaction signals, the emerging model points to a future where assistants learn continually yet remain bounded by clear user choices, a balance that many organizations have sought but struggled to implement. If this direction holds, personalized commerce becomes the proving ground for the next phase of enterprise content, where value grows with every consented interaction.