The agent-native PIM for catalogs.

Your team works in a fast, calm interface. AI agents work through the same 200+ tools, with the same permissions: import, enrich, translate, publish. One source of truth for both.

Explore the features See how it works

Agent edits land as needsReview, not as surprises.

Today, your catalog is a folder of spreadsheets.

Prices in one file, translations in another, images on someone's desktop. Every channel gets its own export, every export gets its own mistakes. Nobody is sure which version is true.

Origo replaces the folder with one structured catalog that every channel and every colleague reads from.

  • products_FINAL_v7(2).xlsxedited by 4 people
  • EAN-list-Q3-DEFINITIVE.csvout of date
  • translations_DE_check-with-Eva.xlsxEva left in March
  • amazon_export_NEW_use-this-one.csvdon't use this one
  • IMG_final/final2/really-final/1,204 files
  • One catalog in Origoalways current

From scattered files to every channel, in four steps.

Product information has a supply chain of its own. Origo gives each stage a home, so data flows forward instead of being copied sideways.

Step 1 of 4

Centralize everything you already have

Import supplier feeds and legacy exports, map columns to real attributes once, and let async jobs churn through the rest. Files land in built-in media storage with folders, usage tracking, and automatic cleanup.

  • CSV import with column-to-attribute mapping and job history
  • Media library on object storage: folders, labels, usage tracking
  • Full REST API for continuous sync from ERP or suppliers
Import supplier-feed-spring.csv
artikelnaam→Name (nl-NL)
ean_code→EAN
breedte_cm→Width (cm)
merk→Brand
2,131 of 2,598 rows4 warnings
Step 2 of 4

Enrich in every language, without the copy-paste

Attributes are typed (text, numbers, booleans, dates, lists, links, measures with units) and multilingual. Locale tabs keep translations side by side, and a review flag marks what still needs a human eye.

  • Ten-plus attribute types, grouped into panels per product family
  • Per-locale values with translation review states
  • Variants inherit from parents; override only what differs
Oak Dining Table 180 TBL-0180-OAK
NameEiken eettafel 180 cm
DescriptionMassief eiken tafelblad met een geolied…
MaterialSolid oak list
Width180 cm · measure
AssemblyRequired de-DE needs review
Step 3 of 4

Structure the catalog the way you actually sell

Hierarchical categories, brands, associations, and parent/child variants keep thousands of products navigable. Classifications define which attributes a product type requires, so "complete" means something.

  • Category trees with per-channel assignments
  • Variant matrices: see coverage across color, size, or any axis
  • Classification templates with required attributes and defaults
Structure
Categories
Variant matrix
Step 4 of 4

Publish to every channel from one place

Channels carry their own locales and categories; listings let a product say something different on each one. A status workflow (draft, prepared, reviewed, ready) keeps half-finished data off your storefronts.

  • Per-channel listings with attribute overrides
  • Product feeds, async exports, and PDF datasheets
  • Completeness at a glance before anything goes live
Channels · Oak Dining Table 180
Webshop nl-NL · en-USListed · all required attributes present 100%
Marketplace DE de-DE2 required attributes missing 86%
B2B portal nl-NLExport queued · runs nightly 64%

Everything in the box.

The full feature list. No modules to license, no add-ons to unlock. One system, one price conversation.

Catalog 01

Products & variants
Parent/child hierarchies with inheritance, variant matrices, duplication, and restore.
Categories
Unlimited tree depth, drag ordering, per-channel category assignment.
Brands & associations
Brand records with their own attributes; cross-sell, up-sell, and custom relation types.
Search & filters
Fast full-catalog search with saved filters across attributes, status, and completeness.
Status workflow
Draft → prepared → reviewed → ready, with a blocked state that's impossible to miss.

Content 02

Typed attributes
Text, numbers, booleans, dates, JSON, single and multi lists, links, and measures with units.
Attribute panels
Group attributes into panels and definition groups so editors see what matters per family.
Classifications
Templates that declare required attributes and defaults per product type and channel.
Multilingual by design
Every field stored per locale — names and descriptions included, not just attributes. BCP 47 locales, fallback chains, translation review states, and safe base-locale switching.
Labels
Lightweight tags for campaigns, batches, and workflows: assign, merge, filter.

Media 03

Built-in DAM
Files on object storage with folders, metadata, and usage tracking across entities.
Asset bindings
Attach, order, and override images per product or variant; set the main image in one click.
Import from URL
Pull assets straight from supplier links; cleanup runs async in the background.

Distribution 04

Channels & listings
Per-channel locales, categories, and attribute overrides for every storefront.
Feeds & exports
Product feeds, async export jobs, and PDF datasheets generated on demand.
Imports at scale
Queued import jobs with mapping, validation, warnings, and full history.
Activation
Push the catalog straight into Shopify, BigCommerce, Channable, or a search index — no middleware in between. Verified destinations, queued publish runs, full run history.

Governance 05

Roles & access
Admin, editor, and viewer roles enforced across UI, API, and agent tools alike.
API keys
Scoped keys for integrations and agents: create, rotate, revoke.
Audit log
Every change records who made it — a colleague in the UI, an integration on the API, or an AI agent. Soft delete everywhere, so deletes can be undone.

Platform 06

REST API + OpenAPI
A documented, versioned API covering everything the UI can do.
MCP server
200+ tools mirroring the full API, ready for Claude and other AI agents.
Edge-native
Runs on a global edge network, close to your users; no servers to patch or scale.

The first PIM your AI agents can actually operate.

Most systems bolt a chatbot onto the side. Origo is built the other way around: every action in the API is mirrored as an MCP tool. Anything a person can do in the interface, an agent can do with a scoped key.

  • Full parity. New endpoint, new tool, same release. Every time.
  • Same rules apply. Agents authenticate with API keys and inherit the same role-based permissions as people.
  • Humans stay in the loop. Agent edits flow through the same review states and activity history as everyone else's.
  • No code editor required. Connect Origo to Claude Desktop or ChatGPT and ask in plain language — "translate the spring collection to German" — straight from the chat window.

Quietly fast, everywhere.

Origo runs serverless on a global edge network. That is why the product list opens as fast in Rotterdam as in Singapore, with no server setup on your side.

No servers to manage
Serverless compute scales from one editor to a full import run without anyone touching a dial.
Media on object storage
Assets live on object storage with delivery close to every visitor.
Async where it counts
Imports, exports, and cleanup run on queues; the UI never waits on a batch job.
Modern auth
Session sign-in for people, scoped API keys for machines and agents, and SSO without the usual enterprise tax.

Common questions.

Something else on your mind? Just ask — we'd rather show than tell.

What is a PIM, and do we need one?

A product information management system is the single place where everything about your products lives: names, descriptions, specs, translations, images, and where each product is sold. If your team maintains more than a few hundred products, sells on more than one channel, or works in more than one language, spreadsheets stop scaling. Usually all three happen at once.

How is Origo different from other PIM systems?

Three things. It's calm: a fast, quiet interface built for people who spend hours a day in it, not for a sales demo. It's agent-native: the entire API is mirrored as MCP tools, so agents work with the same access people have. And it's edge-native: it runs serverless on a global edge network, so it's fast everywhere and there is no infrastructure to set up or maintain.

Can we bring our existing data?

Yes. That's step one, literally. CSV imports map your columns to typed attributes, run as background jobs, and report warnings per row. For continuous sync from an ERP or supplier feeds, the REST API covers every entity in the system.

How do AI agents work with Origo in practice?

You create a scoped API key, connect the Origo MCP server to your agent (Claude, for example), and the agent gets 200+ tools: search products, write attribute values in any locale, attach files, manage listings. Permissions are enforced by role (an agent with editor rights can't administer users), and its edits show up in activity history and translation review like anyone else's.

Which channels can Origo feed?

Any channel that can read a feed, an export, or an API. You model each destination as a channel with its own locales and categories, publish products as listings with channel-specific overrides, and deliver via product feeds, scheduled exports, or direct API pulls.

What does it cost?

Pricing scales with catalog size and channels. No per-module surprises; the feature list above is the product. Get in touch and we'll give you a concrete number.

Curious what your catalog looks like in order?

Origo is built by rb2. Questions, ideas, or a messy export you'd like to see structured — drop us a line.

Get in touch Read the features again