Cross-sell and up-sell assistant
A chat agent that proposes which products to offer together with a product, and adds them to that product after your yes. You give it a SKU or a product name.
This assistant works on one product and uses AI to judge what fits. For proposals for the whole catalogue from what customers buy together, use Cross-sell relations from order history.
Two runs of the same question can give different proposals. The assistant is told to wait for your yes before it adds anything, but an AI can act too early. Check the product afterwards.
| In short | |
|---|---|
| Where you use it | On the Products & Clusters list in the Backoffice, behind the Workflows button |
| Starts | With a chat message |
| Changes in Propeller | Adds relations between products |
| Uses AI | Yes, OpenAI. It gets your messages and the product data the assistant looks up |
| Texts | The language of your message |
| Runs in n8n | One per message |
| Difficulty | Advanced. Expect to tune the instructions for your catalogue |
What you get
The assistant proposes up to three products, each with a kind of relation: Accessories, Alternatives, Options, Parts or Related. After your yes it adds them. You find them on the Storefront tab of the product, where you can also remove them.
To get a set number of proposals, say so, for example Please propose two cross-sell products for SKU ABC-123.
What you need
- An n8n account with the credential
Propeller API key. See Set up n8n. - An API key with Owner access on Product Data. Adding a relation needs Owner access.
- An OpenAI account. You pay OpenAI per use.
- Backoffice access and the Configuration role with Owner access, to register the agent.
How it works
- The assistant finds the product. When several products match a name, it asks which one you mean.
- It collects candidates from the same category and with a search in other categories. It leaves out the relations the product already has.
- It proposes a short list and asks you to confirm.
- It adds what you confirm and tells you the result.
The workflows and agents in the Agent Hub are free to use and adapt. They are examples, not supported product features. Test them in a test environment and check the results before you use them on real customers.
Install it
Install a workflow from the library explains each step.
- Import the file into n8n.
- Open the node OpenAI Chat Model and pick your OpenAI account.
- Check that the four nodes that talk to Propeller use the credential
Propeller API key: get_product, search_products, list_category_products and add_crossupsell. - Follow Register the agent in Propeller with the values below.
- Go to PIM > Products & Clusters, click Workflows, pick the agent and type a SKU.
| Field | Value |
|---|---|
| Trigger | Chat |
| Interaction mode | Multi turn |
| Placements | Back office · Products, the Products & Clusters list |
The assistant answers with a proposal and a question. Test with a product you can check afterwards: after a yes the relations are real.
Make it yours
This example has no Settings node. You change it in three places.
| What | Where |
|---|---|
| How it searches and which kind of relation it picks | The instructions in the node AI Agent |
| The AI model | The node OpenAI Chat Model. A stronger model follows the steps more reliably, a smaller one answers faster |
| Another environment | The address in each of the four nodes that talk to Propeller |
Limits
- An answer took 2 to 5 seconds in tests with the standard model. Propeller waits 30 seconds for an answer by default, so a slower model can run into a timeout.
- The search matches words as they are written. A product with an unusual spelling can be missed.
Related pages
- Create and edit a product - check or remove the relations
- Register an AI agent