Instinct’s credit card adoption shows why AI shopping agents are becoming a major tech race

Instinct founder Noah Shinn said nearly 40% of users connect personal credit cards within three weeks of using the AI personal assistant. He presented the figure at Fireworks AI’s Forge conference in San Francisco as evidence that users are quickly learning to trust AI agents with money and sensitive information. Shinn called it “a real sign of trust.”

Instinct is part of a wave of assistants built to complete tasks, not just respond to prompts. The article lists examples including booking flights, buying groceries, managing calendars, and canceling subscriptions. That shift raises the stakes because more useful agents need access to emails, calendars, payment details, and user preferences.

The race includes Meta’s Muse, OpenAI’s Dots, and Anthropic’s Claude Duet, according to the article. Shopping is emerging as a major use case: Meta has said more than a quarter of Muse users made purchases through the agent, with an average order exceeding $200. Instinct also measures trust by tracking when users allow expensive purchases, including transactions above $500 or trips costing several thousand dollars.

Instinct announced in September that it raised $1 billion at a $10 billion valuation. Shinn said the product’s usefulness depends on accumulating information such as private emails, calendars, conversations, and preferences so the assistant can remember details without repeated prompting. He described it as “a massive amount of data” and also discussed plans for real-time voice interactions and dynamically changing interfaces.