Skip to main content

Content automation

Write product content on import The automation runs without you

AI writes the prose, Content Logic fills the structured lines, and an automation runs both when a product is imported or updated.

  • 36 automation templates
  • 7 variable scopes
  • Test Run before you activate
One pipeline on the canvas: After Import → New Products → Clean Up Data → Generate Titles, Write Descriptions and Generate SEO in parallel → Publish to Store → Slack the team.

Key features

  • AI writes the prose

    Descriptions, short descriptions, titles, SEO meta and image alt text, generated from the attributes a product already carries.

  • Content Logic fills the rest

    Variables and conditional blocks resolve shipping lines, warranty text and channel formatting from product data rather than from a prompt.

  • One canvas runs both

    Drop a trigger, a product filter and the actions onto the Automation Builder canvas, and let it fire on every import.

Three layers

Three layers, stacked in that order

AI handles the prose, Content Logic handles the parts that have to be exact, and the Automation Builder decides when either one runs.

  • AI generation

    A prompt turns the attributes on a product into a description, a short description, titles, SEO meta or image alt text.

    Result: draft prose on every product in the selection.

  • Content Logic

    Variables like {{product.ean}} and conditional blocks resolve to real values the moment a product exports.

    Result: the same structured line on all 10,000 products.

  • Automation Builder

    A trigger, a product filter and a chain of actions on one canvas, saved as a draft until you activate it.

    Result: the next import goes through the pipeline without you.

The full pipeline, step by step

  1. Try it on one product

    Select a product, pick Descriptions and choose a prompt. You see what the AI makes of the attributes you actually have before you commit to a batch.
    One field, one prompt, a few products. Your own prompt is the default; the curated prompts sit underneath.
  2. Tune the prompt until it sounds like you

    Prompt Lab puts the original product and the AI version side by side, with the field, the industry and the tone of voice as inputs. Save the prompt when the output is right.
    Prompt Lab: original left, AI right, and the field you are testing (Description, Titles, SEO Meta, Image Alt Text) as tabs underneath.
  3. Run it over the whole selection

    Bulk Enrich applies a prompt per field to every product in a filtered selection. The product count and the credit cost are on screen before the run starts.
    The bulk run: tick the fields, pick a prompt for each, and the estimate tells you what it costs before you start.
  4. Add the parts AI should not guess

    Content Logic variables resolve from product data (brand, price, SKU, country of origin) so exact values stay exact. Type two braces in any field and the preview shows the resolved sentence.
    The template on top, the resolved sentence underneath. Every variable becomes the product's real value on export.
  5. Wire it into an automation

    One node holds the enrichment action; the canvas holds the trigger, the filter and the publish step. Test run it on a handful of products, then activate.
    An enrichment node: the field it writes, the prompt it uses, and the state it reports back to the canvas.

Automation Builder

Start from a template

New product onboarding, nightly store sync, a translation run. 36 workflows ship ready to open and change.

  • Drag, do not code

    Triggers, conditions and actions are blocks on a canvas. Connecting them is the whole configuration.

  • Draft until you say otherwise

    An automation stays a draft until you activate it, and Test Run executes it on a small set of products first.

  • Triggers that match real events

    After an import, on a schedule, or when a product is created or updated in the catalog.

36 templates, filtered by category and by how long they take to set up. Most of the beginner ones are one to three minutes.

Where teams start

Three pipelines people build first

Each one is a template you open and change rather than a workflow you design.

  • New product onboarding

    An import triggers enrichment, translation, a quality check and a publish to every connected channel.

  • Seasonal refresh

    A scheduled run rewrites promotional lines and pricing labels across one category.

  • A second market

    A translation run plus channel-specific formatting, so a new country does not need a new team.

FAQs

Automation questions, answered

How the three layers fit together, and where each one stops.

Still have questions?

Can't find the answer you're looking for? Please get in touch with our team.

Contact Support

Three things running together: AI generation for prose, Content Logic for content built from product data, and the Automation Builder for the trigger that starts both. Each one works on its own, so you can use AI generation without ever building an automation.

A prompt writes prose and nothing else. Content Logic covers the parts a prompt should not touch, such as the SKU, the EAN, a shipping line derived from weight, or a channel-specific format, because those resolve from the product record. The Automation Builder then decides when either one runs, so a new import does not wait for someone to select it.

Yes, through the prompt. Prompt Lab lets you set industry and tone of voice, run the prompt against a real product, compare the result with the original, and save it when it is right. Saved prompts are reusable in bulk runs and inside automation nodes. The model follows the prompt; it does not learn your brand on its own.

No. Triggers, conditions and actions are blocks you drag onto a canvas and connect. 36 templates cover the common pipelines, grouped by category and by how long they take to set up. There is an API request block for the cases the built-in actions do not cover.

Yes. An automation stays a draft until you activate it, and Test Run executes it against a small set of products so you can read the results first. Nothing runs against the full catalog until you activate the automation yourself.

Quality Guard rules run before the publish step, and a product that fails is held rather than pushed to a channel. You see it in the run's result list and can fix it, re-run that step, or let the automation skip it on the next pass.

Next step

Build one pipeline for your next import

Start free, open a template, and run it against ten products to see what it writes.