Pinterest has launched Ask Pinterest, an experimental app built around conversational, visual-first, and agentic shopping experiences, with the announcement made at Cannes on 17 June 2026.

The app is the sharpest expression yet of a strategic direction the company has been building toward for several years. Pinterest CEO William Ready stated in late 2025 that Pinterest has spent the preceding three years deliberately repositioning itself as an AI-powered shopping assistant, serving close to 600 million users. Ask Pinterest is where that repositioning becomes a distinct product surface rather than a layer embedded in the existing feed.

The pivot from passive discovery to active, conversational commerce is not abrupt. Pinterest introduced Pinterest Assistant in October 2025 as a visual-first shopping collaborator — a signal that the company was already moving beyond static pinboards toward something closer to a guided buying workflow. Ask Pinterest appears to extend that logic, adding an agentic dimension: the app is framed not just as a search or recommendation engine but as a system capable of acting on a user's behalf within a shopping context.

What makes this architecturally interesting to anyone tracking the agentic AI space is the explicit visual-first constraint. Most conversational commerce experiments — from LLM-backed chatbots bolted onto e-commerce storefronts to voice-driven shopping on smart speakers — have treated imagery as a secondary retrieval artefact. Pinterest is betting its core asset, a decade-plus graph of visual intent signals and user taste data, is precisely the thing that gives an agentic shopping experience its utility edge. Text queries alone are notoriously lossy when the user's goal is something like "a lamp that feels like this room" or "shoes that match the mood of this outfit." Multimodal retrieval, grounded in a platform that has trained on billions of user-curated images, is a credible answer to that lossiness.

Worth flagging: the word "experimental" in Pinterest's own characterisation of Ask Pinterest is doing real work here. Cannes is a marketing festival, and announcements timed to it are designed to generate attention from brand partners and agencies as much as from consumers. It is not yet clear what the app's release scope looks like, whether it is a limited beta, a standalone iOS/Android launch, or a web-based sandbox. The agentic framing — systems that act, not just respond — carries significant product and liability complexity that a "experimental" label deliberately leaves unresolved.

Pinterest has form in shipping features that expand the platform's inclusion surface alongside its commerce surface. The body type ranges tool launched in March 2024 let users self-select which body types appear in their search results — an acknowledgment that a shopping-adjacent discovery platform carries real-world weight for how people see themselves reflected in aspirational content. Whether Ask Pinterest inherits or extends those controls matters: an agentic shopping layer that surfaces product recommendations without the same inclusivity safeguards would be a step backward on a dimension the company has publicly committed to.

The Cannes timing also places Ask Pinterest alongside Pinterest's broader brand-partner narrative. The company returned to Coachella in April 2026 with a phone-free experiential activation — an unusual stance for a platform whose entire existence depends on people pointing phones at things. That tension is intentional; it positions Pinterest as a platform that understands aspiration and lived experience, not just scrolling behaviour. Ask Pinterest, pitched to Cannes audiences, is the monetisation counterpart: a product that converts that aspirational intent into transactional outcomes.

The competitive landscape here is dense. Google's Shopping Graph, Amazon's Rufus assistant, and a growing field of AI-native shopping startups are all working the same seam between discovery and purchase. Pinterest's differentiation thesis is that its user base arrives with higher purchase intent than a general search engine and with more curated visual taste data than a marketplace. Whether an agentic layer is enough to hold that differentiation — and whether the "experimental" tag becomes a full product on a competitive timeline — are the questions worth watching through the rest of 2026.