An example setup for an e-commerce or marketplace team that manages its own catalogue: catalogue work moves to the night, the decision to publish stays with you.
It runs on your own server: work files and records are stored on your infrastructure; content processed by AI goes to the model provider you choose, and data sent to external tools goes to the connections you set up.
Chat · catalogue
demo data
Write descriptions for the products added this week, in our brand language. Send the ones I approve in the morning to the shop.
18:30
Remyral18:30
I prepared a workflow draft that runs every evening at 22:00: the loop handles 4 products at a time, and every description is checked against the schema and a 160-character limit.
Approval · 07:45Awaiting approvalApproved
48 product descriptions
Bamboo chopping board · 3 bullets · 152 chars
Ceramic mug 350 ml · 4 bullets · 148 chars
Linen tablecloth · 3 bullets · 157 chars
Workflow resultCompleted
45 products in the shop
To edit
3 products
Output
descriptions.csv
01Workflowfrom the chat or the editor
Productdescriptions
Built from parts: new products are pulled, a loop runs one AI step per product, the output is validated against a schema, and in the morning the approved ones go to the shop.
Workflow · Product descriptions
demo data
Product descriptionsSavedSaveRun
Last run · 22:00 · 48 products · 45 approved
02Setup and flowE-commerce catalogue
Setup and flow
An e-commerce or marketplace operations team that manages its own product catalogue.
Setup
Project type
A workspace project; workflow engine 2 switched on.
Input
A CSV exported from the shop panel or an HTTP request to the shop API; a large catalogue is fetched page by page.
Brand language
Tone, phrases to avoid and character limits go into memory as rules.
Lightest setup
It also works without an API: network off, files in, files out. To write to the shop, the API key is in the vault, bound to the shop’s address.
What a day looks like
01
In the evening you put the product list into the workspace, or the workflow pulls new products from the shop API; a CSV step parses the list.
02
A loop runs one AI step per product (1–20 at a time, 4 by default): title, description, bullet points and meta are validated against an output schema and filtered by character limit.
03
In the morning you read them on the approval screen; approved items go to the shop API. Everything also stays in the Outputs screen as descriptions.csv, shown as a table.
What changes
Catalogue work moves to the night; the decision to publish stays with you. An idempotency key keeps a request from being sent twice; on a large run a budget cap limits the spend.
How it connects
Instead of a ready-made connector, you connect to your shop’s or marketplace’s API with workflows and automations and set the fields to fit your own setup. If you’d rather not use an API, you start from an exported CSV and get CSV back.
We don’t impose a ready-made flow; design your own, either with AI (describe it in the chat) or yourself (in the workflow editor).
Good to know
Flows such as stock, orders or returns can be built from the same parts as workflows of their own.
Outputs are delivered as CSV, Markdown or JSON; CSV opens as a table in the Outputs screen and imports into Excel.
Job · live
demo data
A screen from the console · The real console: the job page, with live log and timeline.
This example runs on the same infrastructure as all the others: Remyral on your own server, and we build your flow together in the pilot. Just describe the work you want to simplify and the tools you use in your application.
Remyral18:30
