Pinterest Launches Ask Pinterest, a Standalone AI App for Conversational Shopping Discovery

Quick Facts

Ask Pinterest is a standalone experimental app that processes natural-language queries and delivers personalized shopping recommendations using Pinterest’s Taste Graph.
Pinterest reported $1.008 billion in Q1 2026 revenue, up 18% year over year, with 631 million monthly active users.
The company also unveiled three companion advertiser tools: Business Assistant, Pinterest MCP, and an expanded Performance+ creative model.

Pinterest launched an experimental AI-powered app called Ask Pinterest on Wednesday, targeting users making complex, multi-step decisions that a standard search bar cannot handle. The company announced the app ahead of the Cannes Lions advertising festival, where AI is expected to dominate the agenda.

The app is currently available to a limited number of users in the United States on mobile and desktop web. Pinterest says it will serve as a testing ground for AI features that may eventually appear in the main Pinterest app.

How It Works

Ask Pinterest uses a conversational chatbot interface. Users type natural-language questions and receive product recommendations drawn from their saved Pins, Boards, and past activity. The app retains context across sessions, allowing for follow-up questions without losing prior conversational history.

The app runs on Pinterest’s proprietary Taste Graph, an internal system that maps user preferences and aesthetics based on platform activity across its 631 million monthly active users. Pinterest is not licensing its recommendation data to third-party AI services. It is building these models in-house.

“Ask Pinterest is a new experimental app for exploring more conversational, visual-first and agentic shopping experiences,” the company said in a statement Wednesday.

Advertiser Tools Announced Alongside the App

Pinterest also announced three tools aimed at advertisers. Business Assistant is an AI tool inside Pinterest Ads Manager that analyzes campaign performance and surfaces recommendations. It is currently in closed beta with a limited group of advertising partners.

Pinterest MCP, a Model Context Protocol, connects Pinterest campaign data, analytics, and keyword insights to third-party AI tools already used by agencies. Alpha partners include PMG, Pacvue, Dentsu, Havas, Innovid by Mediaocean, and Omnicom’s Jump450. The approach follows similar moves by TikTok and Meta, both of which opened their ad platforms to third-party AI agents in 2026.

The company also expanded its Performance+ creative tool with a new AI model that selects the most relevant creative variation for each individual ad impression. Internal testing showed a 7.5% increase in click volume compared with the previous model. The feature is now available globally.

Financial Context

Pinterest posted $1.008 billion in Q1 2026 revenue, an 18% year-over-year increase. The company reported a record $4.2 billion in full-year 2025 revenue, up 16%. Monthly active users grew 11% to 631 million in Q1 2026.

Clicks on advertiser links grew 40% in the most recent quarter compared to the prior year and have increased more than five times over the past three years, according to the company’s earnings reports.

CEO Bill Ready told analysts that Gen Z now makes up more than half of Pinterest’s user base. The company signed a $4 billion cloud infrastructure commitment with AWS through 2031, the largest infrastructure deal in its history, to support AI model training and platform growth.

What This Means for Retailers

Ask Pinterest signals a direct push into AI-powered shopping discovery. Brands already active on Pinterest should watch how the Taste Graph shapes product recommendations inside the new app, since those signals could influence what surfaces in the main platform over time.

The MCP partnerships with major agencies suggest Pinterest is building infrastructure for programmatic AI-driven ad buying. Retailers working with Dentsu, Havas, or Omnicom may gain early access to these tools before a wider rollout.