Skip to main content

What are AI Agents and how they work

An introduction to AI Agents and the four categories in AI Studio

Written by Alberto Barberis

An AI Agents is a specific task the AI performs on a single product attribute. Each agent targets one attribute and defines precisely how the AI should generate or transform content for it.

AI Studio page showing the Actions Library with four categories and pre-built action cards

AI Studio — the Agents Library on the left, pre-built actions and Create Agents cards on the right.

AI Agents are organized into four categories:

  • Content Generation: Title, Description, Short description, Attributes variables, Bullet points, FAQ, Translation

  • Product Organization: Smart attribute, Category, Tags

  • SEO Optimization: Page title, Page meta description, Amazon keywords, SEO Keywords

  • Data Enrichment: Enrich Product

There are two types of actions:

  • Pre-built Agents — shown as purple cards with a "By AndromedAI" badge. These work immediately with no configuration required. They represent the standard approach for each content type.

  • Custom Agents — built by you using the agent editor. You define specific behavior, character limits, output format, glossaries, banned words, and custom rules tailored to your brand or product category.

Each agent has an estimated duration shown when you select it in the Optimize workflow, helping you plan before running a large batch.

Agents consume credits when run at scale. Test Agent lets you preview results on 1–2 sample products before committing to a full batch run, at no extra cost.

Did this answer your question?