What we’re shipping in Q4 and what comes next in Q1 2027

Shipping in Q4 2026

Product groups. All the colors and versions of a style stay side by side in a collection, even when the sorting is automatic.
A/B testing. Run two versions of a collection, with different sorting, rules or layout, and see which one wins on click-through and the other numbers you follow.
Segments. Logged-in, returning or tagged shoppers each see their own version of a collection.
Looks. Put a set of products together, like an outfit, and show it as one in collections, search and recommendations.

Rules per search term. For a specific search, pin or boost products, add or remove them, or send the shopper straight to a collection. Synonyms stay automatic.
Collections in search. A search for dresses also shows your Dresses collection.
Autocomplete. Suggestions while the shopper types.
Custom filters. Filter on any product attribute, like material or heel height.

Recommendations you run yourself. Set rules, pins and looks, then preview the result. No developer needed.
Recommendations that follow the visit. Recently viewed, similar to what the shopper just looked at, and add-ons in the cart. Live now.
A/B testing. Compare two recommendation setups.

Depict in your AI assistant. Connect Claude or ChatGPT, ask how your store performs and change your merchandising in the chat. Live now.
Insights you can act on. Reports that tell you what to do next, like a product that gets seen but not clicked, and make the fix for you.
A faster API. The new real-time API runs closer to the shopper, answers faster and is simpler to integrate.
AI shopping assistants. More shoppers start with an assistant like ChatGPT. We are working on how those assistants find products in your store, and we will share more when it is ready.

Smart sorting. Today sorting rules stack in layers, all the new arrivals first, then all the bestsellers. Smart sorting gives every product one score across sales, newness and stock, so a product that is both new and selling gets credit for both. Depict picks the starting mix. You tune the weights, see why a product sits where it does, and test it against the sorting you have today.
Social blocks. Instagram, TikTok and Pinterest posts placed in the product grid like any other content block.
Rules per market. Pick one, several or all markets for a campaign or a rule, or copy it to other markets without rebuilding it.
Content blocks. Choose the shape and height of a tile, add translated text or a button in your store’s font, and build a tile once to reuse across collections and markets.

Attribute search. Depict reads size, stock and price from the search itself, so “black boots size 10 under $45” shows only matching products that are in stock.
Quicklinks. Chips above the results that narrow them in one tap, like Men’s and Women’s for “t-shirt”.
AI overviews. A short AI-written summary above the results.

Complete the look. AI picks the items that go with a product, and you can hand-pick your own.
Automatic A/B testing. Depict tests recommendation setups for you and shows you the results.

Each feature launches on one platform first and then rolls out to every setup. Every one gets its own update here the day it is live for you.