Prada Enters ChatGPT. The Brand Now Has to Work Without the Page

A luxury website decides what you see first. The campaign image, the collection, the silhouette, then perhaps the price. That order is part of what the brand is selling. Prada is now testing an experience in which the customer can start anywhere: with a material, an occasion, a product or the location of a store. Its new ChatGPT pilot draws on Prada's own brand sources, with Accenture Song and DaVinci Commerce behind the project. The announcement is about product discovery. The more consequential question is what happens to a carefully composed brand when its content has to answer questions in an order nobody in the creative team chose.

Key takeaways

• Prada's ChatGPT pilot lets users explore collections, discover products and find store information in the US, UK and Italy, using information grounded in the brand's own sources.
• A branded conversational experience gives a company a way to supply the material behind an interaction. It does not establish control over every answer ChatGPT gives about that company.
• For creative teams, the work extends beyond producing approved pages. Product facts, brand language and the context connecting them need to remain coherent when retrieved separately.

What Prada actually launched

Coverage published on 28 October describes a pilot developed by Prada and Accenture Song, with support from DaVinci Commerce. Users can explore the house's collections, discover products and access store information through conversations in ChatGPT. The experience is available in three markets: the United States, the United Kingdom and Italy.

The announcement describes discovery and engagement. It does not establish that customers can complete a purchase inside the experience, and it publishes no conversion results. Calling it a fully operational store would get ahead of the evidence.

The detail that matters is the source material. The experience is grounded in Prada's own brand sources. That creates a different starting point from an assistant assembling a description out of whatever it retrieves elsewhere.

For this pilot, the house is participating in how it is represented. What that participation can achieve is the question worth examining.

The page used to hold the argument together

Consider how a collection page works. Photography establishes an atmosphere. Styling places a piece in a particular world. Copy explains a reference or a technique. The product details arrive inside that setting.

Those elements do different jobs, but their proximity lets them support one another. A short description can rely on an image for context. A material can acquire significance from the story placed beside it. The customer may not read everything, but the brand has composed the encounter.

A conversation changes the sequence. Someone can ask whether an item suits a particular occasion before seeing the collection. Another person can begin with dimensions. A third can ask why two apparently similar products differ in price.

These are examples of the questions a team should prepare for, not claims about functions tested in Prada's pilot. They illustrate the production problem: information written to sit inside a page may be asked to work on its own.

The creative task becomes deciding which context must travel with it.

Accurate information is only the beginning

A correct answer can still be an inadequate representation of a brand.

Two products may share a material, a colour and a category while differing in construction, design intent or their relationship to a collection. If the available information describes only the shared attributes, an assistant has little basis for explaining the distinction.

The answer may be factually clean and commercially generic.

This is where the subject reaches beyond ecommerce feeds. Creative teams produce the language that makes differences intelligible: why a detail exists, what a collection is exploring, how a particular technique contributes to the object. That material needs enough specificity to be useful in a response, rather than a string of adjectives that could describe any competitor.

The operational implication is straightforward. Brand expression and product information need to be developed together. Neither can reliably compensate for gaps in the other once the customer starts asking precise questions.

The agency brief extends beyond launch

The partners behind the pilot help explain the kind of work this creates. Accenture announced its investment in DaVinci Commerce in March, alongside a strategic partnership with Accenture Song. Its stated ambition spans discovery, merchandising and the wider commerce journey, with brand assets becoming material for experiences in AI environments.

For a creative organisation, that suggests a brief with a longer life than the launch deliverable.

A campaign can be approved and published. A conversational experience also needs a maintained body of information: which description is current, which market it applies to, what supports a claim and what should happen when the source cannot answer a question.

Someone has to own those decisions after the experience goes live. A discontinued product, revised description or new collection can change what a previously acceptable response should say.

This is an implication of operating such a service, not a description of Prada's internal arrangements, which the launch coverage does not disclose. It is also a concrete opportunity for agencies: designing the experience and maintaining the content that makes it credible belong in the same conversation with the client.

Owning the sources is not owning the answer

There are two different ambitions here, and they should not be confused.

One is to build a useful branded experience for people who choose to enter it. The other is to make sure a general-purpose assistant recommends the brand whenever someone asks about the category.

The pilot addresses the first. Its existence does not demonstrate the second.

A customer already exploring Prada and a customer asking which bag to buy are at different points in their journey. The former may welcome deeper access to the house's information. The latter may never encounter its dedicated experience. A branded presence therefore needs its own discovery and measurement plan.

Source control has limits inside the experience too. Grounding responses in brand material is a meaningful design choice, but the announcement is not an independent evaluation of answer accuracy. It does not explain every retrieval rule, update process or safeguard against unsupported claims.

The useful distinction is between supplying authoritative material and proving that the resulting interaction consistently uses it well.

What this pilot cannot yet prove

There is no published evidence in the launch reports that this experience increases sales, attracts new customers or improves the quality of advice compared with Prada's existing channels.

Nor does entering ChatGPT establish that conversation is the best format for every part of luxury discovery. A customer may want to browse images, understand a silhouette through movement or encounter a collection without having to formulate a question first.

That leaves room for a more modest, and more useful, interpretation. Prada is testing an additional way into its universe. The experiment could identify which questions conversation handles well and where a website, a campaign or a person remains more effective.

For brands considering a similar move, those boundaries should be part of the brief. Otherwise a launch can be called successful because the interface exists, without anyone establishing what it does better.

Three decisions before commissioning the experience

First, decide what the conversation is for. Collection discovery, product comparison and store information are different jobs. Pick the intended journeys and define what a useful outcome looks like for each. A qualified visit to a boutique and a completed online purchase should not be forced into the same success measure.

Second, review the source material outside its original layout. Take a product description away from its campaign page. Check whether it still explains the object, supports its claims and preserves the distinctions that matter to the brand. Attach the relevant market, version and approval context. Missing information should lead to a clarification or a handoff, rather than an invented answer.

Third, give response quality an owner. Test representative questions, including ambiguous ones and questions the content cannot answer. Review factual accuracy and brand expression separately. Define who can correct the source and how the correction reaches the experience. These are operating responsibilities, not a final round of copy approval.

This is where the work returns to creative operations. The image, the description and the context that makes them usable need to remain connected as they move between teams and destinations. MTM's role sits in that upstream workflow: keeping creative assets and their associated information organised, so teams can find and maintain the material on which downstream experiences depend.

Prada's pilot puts a practical question in front of every brand team considering the same move. When a customer skips the page and asks a question instead, what have you given the system that makes the answer recognisably yours?

See how MTM helps teams organise creative assets and their associated information →

FAQ

What does Prada's ChatGPT experience do?

The announced pilot supports collection exploration, product discovery and store information through conversation. It was developed with Accenture Song and support from DaVinci Commerce, and is available in the US, UK and Italy. The launch reports describe responses grounded in Prada's brand sources.

Can customers buy Prada products directly inside ChatGPT?

The sources reviewed for this article do not confirm an integrated checkout. They describe discovery and engagement. A wider ambition around agentic commerce should not be treated as evidence that this particular pilot completes transactions.

Does a dedicated experience make ChatGPT recommend the brand more often?

The announcement provides no evidence of that. A branded experience and visibility in general category recommendations are separate questions. Brands should measure how people find the experience as well as what they do once inside it.

What should a creative team prepare before building something similar?

Start with the questions the experience is meant to answer. Check the completeness and consistency of the approved product information, the brand context needed to interpret it and the responsibilities for keeping it current. Then test whether answers preserve both accuracy and meaningful distinctions between products.

Does conversational discovery make visual storytelling less important?

The pilot does not establish that. Visual storytelling can create interest and convey things a factual response cannot. The practical task is to connect that work with the information a customer needs when asking a specific question, and to recognise when conversation should hand the customer to another format or a person.

Sources

• Retail Gazette, Prada launches ChatGPT tool to help shoppers discover luxury products, 28 October 2026. https://www.retailgazette.co.uk/blog/2026/10/prada-chatgpt-tool/
• Retail Times, Accenture helps Prada bring product discovery to ChatGPT, 28 October 2026. https://retailtimes.co.uk/accenture-helps-prada-bring-product-discovery-to-chatgpt/
• Accenture, Accenture Invests in DaVinci Commerce to Advance Agentic AI-Led Shopping, 23 March 2026. https://newsroom.accenture.com/news/2026/accenture-invests-in-davinci-commerce-to-advance-agentic-ai-led-shopping