One product model. Six agents working against it.
Catalog AI is not six tools that happen to share a login. The agents run against the same products, the same attribute definitions, and the same credit balance — which is why running validation over what enrichment just wrote costs nothing to set up.
Five layers, top to bottom
- Your catalog
- Products, brands, categories, and catalogs, with parent-child variant families. Data arrives by CSV, by API, or extracted from files.
- Your specification
- Attribute Sets define what a finished product, brand, or category must carry — the fields, their types, and the prompt behind each one, scoped per category and per brand.
- The agents
- Six jobs that run against that specification: enrichment, validation, market analysis, import, brand, and category. Each with a model tier you choose per run.
- The evidence
- Sources attached to values, confidence scores attached to checks, full history attached to products, and a cost estimate attached to every job before it starts.
- The way out
- Review in the UI, export CSV, bulk-export images, or read everything back through the REST API with webhooks on completion.
Each one answers a different question
They can run independently, or in sequence — import, enrich, validate is the path most catalogs take.
Product Enrichment
Product Validation
Market Analysis
AI Import
Brand Enrichment
Category Enrichment
The unglamorous half.
None of this is why anyone signs up. All of it is why teams stay once the catalog is large enough to be someone's job.
Catalogs
Group products into collections — by supplier, brand, season, or project — and target jobs at one collection at a time.
Variant families
Parent-child products with aggregation, so a family is enriched once and the variants inherit what they share.
Bulk CSV in
Import products, and update existing ones in place matched on internal ID, external ID, or SKU.
Bulk export out
Export jobs for product data and image export jobs for generated and sourced imagery.
Job history
Every run stored with inputs, model tier, cost, and results, per agent.
Organizations
Separate workspaces with their own products, settings, credits, and members, switchable from one login.
Team roles
Invite members with roles. Seats are free — credits are the only thing metered.
API keys
Scoped keys with configurable rate limits and auto-refill, plus usage tracking per key.
Chosen per job, not per plan.
Every agent takes a tier. The same account can run a hundred thousand products on Budget and five hundred on Ultra in the same afternoon.
See what a credit buys- Budget
- Bulk backfills and large low-margin rangesLowest cost per product
- Pro
- Production jobs. The one most teams run.About 1 credit per product
- Ultra
- Ambiguous specs, technical categories, high-value SKUsHighest cost per product
The whole platform is on the free tier.
No feature gates. The free credits run the same agents at the same tiers as a paid account — there is just less of them.
5 credits on signup · no card required