To create a product collection for an AI agent in AgenticCart, you pick a clear purpose, add products from your synced catalog, review the quality, and assign the collection to an agent. This tutorial walks through the full flow so you can build AI shopping collections that actually produce sharper recommendations.
Before you create a collection
- Your store or feed should be connected. If it is not, start with collections and the connect-store guide.
- Your catalog should have completed at least one catalog sync so current products show up in the product picker.
- Decide the job of this collection before you start adding products: category advisor, campaign set, gift guide, bestsellers, or another focused buying journey.
- Know which AI sales agent you plan to assign the collection to, so you can match it to the right tone and objective.
Create a product collection for your AI agent (step-by-step)
-
1
Open Collections in the dashboard
From the left navigation in AgenticCart, go to the Collections area. This is where all of your curated product sets live. -
2
Create a new collection
Click New collection and give it a clear, descriptive name (e.g. "Gifts under 100", "Plant-based protein", "Spring 2025 campaign"). Add a short internal description so other team members know what the collection is for. -
3
Search your catalog
Use the search and filter controls to find products that belong in this collection. Filter by category, availability, price range, or keyword to narrow quickly. -
4
Add products to the collection
Select individual products, or add groups of products that fit the buying journey. Add everything that genuinely belongs — AgenticCart handles collections of any size, from a handful of bestsellers to thousands of SKUs in a full category. -
5
Review and assign to an agent
Before saving, scan the collection for off-topic or poorly described products and remove them. Then open the relevant agent and assign this collection as its recommendation pool. See customize your AI shopping assistant to tune tone, persona, and card layout for the agent that will use it.
Tips for better collections
- Start narrow. A supplements brand's "Sleep support" collection will outperform a general "Wellness" collection for the same query.
- Use products with strong descriptions and images. Thin descriptions handicap the match quality — see product data quality.
- Remove products that do not fit the agent's job, even if they technically belong to the same category.
- Split broad categories into multiple collections when shopper intent differs — a clothing store can run separate "Workwear" and "Weekend" collections instead of one mega "Men's" collection.
- Refresh your catalog before adding newly launched products, so they appear in the picker.
TIP
Focused beats broad. A smaller curated product set with a clear purpose almost always outperforms a large mixed collection, because the AI agent has less noise to cut through when matching shopper intent.
Common mistakes to avoid
A few patterns consistently produce weaker AI shopping collections:
- Too broad a collection. Adding "everything in the category" dilutes recommendations. Filter to the products that actually fit the agent's buying journey.
- Mixed price points. A collection that spans 20 to 2,000 forces the agent to guess which price band the shopper wants. Split by tier if the range is wide.
- Stale descriptions. Products with one-line descriptions give the agent almost nothing to match on. Refresh the underlying product data before adding them.
- Duplicate variants. If the same product appears as multiple SKUs in your feed, pick one and remove the rest from the collection view.
Editing a collection
You can add or remove products any time your merchandising strategy changes. Update collections before campaigns, seasonal launches, inventory changes, or homepage experiments. Edits take effect for new shopper conversations immediately — there is no redeploy step.
Deleting a collection safely
Before deleting a collection, check whether any agent depends on it. If a live agent is assigned to the collection, create or assign another collection first, then delete. This keeps shoppers from hitting an agent with an empty recommendation pool.
Frequently asked questions
How do I choose products for an AI agent collection?
Start from the buying journey, not the catalog tree. Write down the question a shopper would ask this agent, then pick products that could plausibly be the right answer. Products with strong titles, descriptions, and current availability will produce the sharpest recommendations.
Can I import a WooCommerce category directly as a collection?
You can filter by WooCommerce category in the product picker and add those products in a single step. However, we recommend reviewing the imported set and trimming anything that does not fit the agent's specific job — a category and a collection are not always the same thing.
What happens to a collection when products sell out?
Out-of-stock products are automatically excluded from AI recommendations after a catalog sync. The product still shows in the collection view so you can decide whether to remove it permanently or wait for restock. See removing unavailable products for the full lifecycle.
How many collections can one agent use?
Each AI sales agent is assigned exactly one product collection. If you need an agent to handle several distinct buying journeys, build a broader collection; if the journeys are very different, create separate agents instead so each one stays focused.