TechCrunch: Pinterest launches Ask Pinterest, an experimental conversational shopping app

Pinterest announced Ask Pinterest on June 17, 2026 — a standalone experimental app that lets users pose natural language requests and receive personalized shopping recommendations. The app runs separately from the main Pinterest product, a deliberate decision that allows the company to test a fundamentally different interaction model without touching current engagement metrics.

Ask Pinterest is powered by Pinterest’s internal Taste Graph, a database built from billions of saved Pins that maps user interests and aesthetic preferences. When signed in, the app draws on each user’s boards and saves to tailor responses. Unlike a traditional search input, the interface handles multi-step queries: users can ask something like “help me furnish a living room for under $2,000” or “plan a Mediterranean dinner party for twelve” and receive contextual, personalized guidance rather than a grid of keyword matches.

The launch places Pinterest in a growing field of companies building conversational shopping layers over existing inventory and social graphs. Google, Meta, and Shopify have all moved in this direction. Pinterest’s position differs in that its Taste Graph captures aesthetic and intent signals — not purchase history or social connections — which the company argues produces more personalized recommendations for inspiration-driven purchases.

For product managers, the standalone app decision is worth studying. It is a well-established pattern for testing interaction paradigms that may cannibalize or conflict with an existing product: isolate the experiment, recruit a self-selected early adopter base, and collect behavior data before deciding whether to integrate or shut down. The tradeoff is that adoption starts from zero rather than from the installed base, which slows the signal. The benefit is that any success or failure is attributable to the new modality rather than to distribution effects from the main app.

The Taste Graph point also raises a broader product strategy question. Pinterest’s AI layer is differentiated by a proprietary data asset that competitors cannot replicate. For teams evaluating conversational AI features, the analogous question is what unique data or context their product already holds that a generic model does not — and whether that asset is being used in the AI layer at all.