Stripe · AI product · Senior · 45 to 60 minutes

What should Stripe build for agent buying?

Stripe provides payments and financial infrastructure that businesses use to accept money, manage transactions, and build online commerce flows. Answer this product management case and get it graded line by line against a weighted rubric.

product strategy
platform thinking
sequencing
risk management
stakeholder judgment

The situation

You are a senior PM at Stripe. Inbound demand is rising from two sides at once: large sellers want to be discoverable and purchasable through AI agents, and AI platforms want a clean way to complete purchases through Stripe-backed businesses. Leadership believes this could become a meaningful new commerce surface, but it is still early and messy. Sellers have fragmented product catalogs, live inventory that changes constantly, legacy fraud systems built for human web traffic, and support flows that assume a person is on the merchant site. Protocols for agent transactions are also still evolving, and no one knows which standards will matter a year from now. Your VP asks for a recommendation on what Stripe should build first for this space over the next 12 months. The hard part is scope. A narrow payments-only product would be faster to ship, but may not solve the main seller pain points. A broader commerce layer could create more value, but increases complexity and adoption risk. You need to define the product bet, what to launch first, who to launch with, and how to avoid locking Stripe or its users into brittle integrations.

What to produce

Write a 2 page product strategy memo for Stripe leadership.

Your answer must

State the core user problems for sellers, AI agents, and buyers, and identify which one Stripe should prioritize first.
Recommend Stripe's product scope for the first 12 months, including what is in the initial offering and what is explicitly out of scope.
Propose a phased rollout plan, including the types of sellers, products, and use cases you would start with first.
Explain your approach to standards and integrations, including how Stripe should handle multiple agent protocols and seller system complexity.
Define success metrics, leading indicators, and the biggest operational or trust risks, with mitigations for each.

Constraints

You cannot assume a single agent protocol or standard will win.
You cannot require sellers to build a custom integration for every agent surface.
Stripe will often have limited frontend identity and fewer classic browser based fraud signals than in normal checkout flows.
Early use cases should avoid complex fulfillment and support flows where possible.

Format

Use sections for problem framing, recommendation, scope, rollout, and metrics and risks.

How Stripe actually played it

After six months of launching the Agentic Commerce Protocol and working with sellers such as Etsy and URBN, Stripe lists ten lessons on making AI-agent purchases work with real catalogs, inventory, fraud and support. The post is useful for PMs thinking about how a new buying channel hits operational edge cases.

New channels expose edge cases in inventory, fraud and support.
Open standards help when many parties must integrate.
Launch with real sellers to learn the long tail of issues.

Built from 10 things we learned building for the first generation of agentic commerce by Stripe.