Catalog AI
Agent · Category Enrichment

Four hundred category pages, each one written to rank.

Structured category descriptions, meta titles, meta descriptions, and clean slugs across the whole taxonomy — researched per category rather than templated from the name.

Start freeBook a demo

Category · CAT-0431

Angle grinders

Enriched
DescriptionHTML · 312 words, 3 subsections
CoversDisc size · power · guard type · corded vs cordless
Meta titleAngle Grinders — 4½", 5" & Cordless | 58 chars
Slugangle-grindersExisting slug kept — no redirect needed.
Products96 in this category
The situation

Your highest-intent pages are your emptiest ones.

Someone searching for a category is further down the funnel than someone searching for advice and further up than someone searching a part number. It is the best traffic a catalog gets.

And it lands on a grid of thumbnails with a heading and nothing else, because writing four hundred category descriptions was quoted at forty dollars each and quietly dropped from the budget.

So the category ranks below three affiliate blogs that do not sell the product at all.

Output

What each category gets

Description
HTML, 150–350 words: a short opening paragraph followed by two to four subsections covering what the category is, who it is for, what to consider when buying, and the types available.
Meta title
Search-length, carrying the category term without stuffing it.
Meta description
150–160 characters, written for the click.
URL slug
Lowercase, hyphenated, three to five meaningful words, matching your existing singular or plural convention. An existing slug is preferred over a new one, so live URLs do not move.
Custom attributes
Any additional field your category template renders — buying guide, size chart intro, care instructions.
Process

How a category job runs

  1. STEP 1

    Bring the taxonomy in

    Categories can be created directly, imported, or built from the structure you already have.

  2. STEP 2

    Set the specification

    Category attribute sets define the fields and the prompt behind each, so a clothing branch can be briefed differently from a components branch.

  3. STEP 3

    Run the branch

    Select categories — one, a branch, or the whole tree — pick a model tier, and check the estimate before it runs.

  4. STEP 4

    Review and publish

    Read the copy, adjust the prompt if a whole branch reads wrong, re-run, then export or pull it through the API.

Written to be read

Constraints that keep it useful

Most generated category copy is bad in predictable ways. These are the rules that stop it.

Buying considerations, not adjectives

The brief asks for what a buyer actually weighs up in that category — disc size, thread pitch, fit, certification — rather than a paragraph about quality and value.

Related terms, not repetition

Five to ten genuinely related terms drawn from what ranks, instead of the category name repeated until it triggers a filter.

Clean, constrained HTML

No H1, no anchors, no images, no inline styles. It slots into a template that already has its own structure.
Category enrichment

Questions

Will this change our existing URLs?
Not unless you want it to. Where a slug already exists it is preferred verbatim; new slugs are only generated where there is nothing to keep. Moving live URLs is your decision, not a side effect.
Does the copy read like filler?
The shipped prompt bans the usual tells — no "welcome to our range of", no thesaurus words, no repeating the category name ten times — and asks for real buying considerations instead. If it still is not your voice, the prompt is editable.
How does it know what belongs in a category?
It researches the category rather than paraphrasing the name, so the buying considerations reflect what people actually weigh up. It also has your products under that category as context.
Can we do four hundred categories in one job?
Yes. Bulk runs process in parallel with live progress and job history, the same as every other agent.
What about HTML we do not want?
The output is constrained: no H1, no anchors, no images, no scripts, no inline styles. It is written to drop into a template that already owns its own heading and layout.
The rest of the platform

Six agents, one catalog

Product Enrichment

What is this product?Fill every attribute from live sources.

Product Validation

Is any of this still true?Score the data you already have, 0–100.

Market Analysis

Who else sells it, and for how much?Competitor price, MSRP, and stock status.

AI Import

How do I get this catalog in?Supplier PDFs and spreadsheets into products.

Brand Enrichment

Who makes it?Brand profiles written at catalog scale.

Start with the branch that gets the most traffic and the least copy.

One branch is enough to judge the voice. Adjust the prompt, re-run, then let it loose on the tree.

Start freeBook a demo

5 credits on signup · no card required