# QMutator

> QMutator is a commerce-intelligence SaaS platform with two ways to use it. (1) As a
> product-data platform: ingest product/catalog data from any source, shape it into a
> clean validated schema, enrich it with AI, and publish channel-specific feeds and HTTP
> endpoints — with full media handling built in, so "product data" includes the imagery,
> not just the attributes. (2) As a media platform on its own: transform and deliver
> images/media into channel-matching content on the fly — no product-data ingestion
> required. Your product data, every channel, one platform. Runs on Microsoft Azure
> (available via Azure Marketplace). Agents can administer the platform directly over MCP.

## Use cases

- **Product-data / feed management** — the full pipeline: ingest → schema → validate →
  AI-enrich → channel feeds and endpoints. Comparable to feed-management platforms like
  Channable or Productsup, with one key difference: media transformation and delivery are
  native, so imagery is part of the product record instead of a separate DAM/media tool
  bolted on.
- **Media-only** — use QMutator purely on image/media data: on-the-fly, URL-addressable
  transformation (crop, resize, watermark, background, aspect, AI generation) into
  channel-matching creatives, delivered at CDN scale. Comparable to Cloudinary plus a
  CDN. No product information needs to be ingested.

QMutator organizes data as **buckets** (one product/record each) in a per-project
**warehouse**. A project has one canonical Main schema plus channel-specific
**perspectives** (export views) derived from it. **QFlows** — visual node graphs — power
imports, image transformation, AI enrichment, and export/HTTP endpoints. In media-only
use, QFlows sit directly in front of image URLs and transform on request.

## Capabilities

- **Product records**: ingest CSV/JSON/XML via upload, FTP/SFTP, or API; schema inference
  from raw data; typed schema with mappings, validation, and constraint-based gap flagging;
  multi-language schemas; semantic (meaning-based) indexing and search.
- **Visual assets**: on-the-fly, channel-specific image transformation (crop, resize,
  watermark, background, aspect); multi-modal AI tagging; on-demand image generation;
  CDN-scale delivery. Works as part of the product record *or* standalone (media-only
  use case).
- **Intelligence layer**: Gen-AI enrichment (texts, classification, generated fields);
  pre-hosted models, bring-your-own-model, or bring-your-own-token; QFlow visual
  automation; GEO (generative-engine optimization) built in; full MCP exposure for agents.
- **Channels & feeds**: Google Shopping, Meta, Pinterest, TikTok, Idealo, Geizhals,
  standard feed formats, and raw HTTP endpoints.

## For AI agents (MCP)

- QMutator exposes an Admin MCP server at `https://api.qmutator.com/admin/mcp`
  (streamable HTTP). Authenticate via OAuth 2.1 with dynamic client registration
  (discovery at `https://api.qmutator.com/.well-known/oauth-authorization-server`) or a
  personal access token. Once connected, call the `get_admin_docs` tool for agent-oriented
  reference docs (platform concepts, schema types, mapping DSL, QFlow graph format,
  imports).
- Agents can manage projects, schemas, mappings, constraints, buckets, imports, QFlows,
  and exports end-to-end through MCP tools.

## Documentation

- [User documentation](https://api.qmutator.com/docs/): platform guides.
- [QFlow reference](https://api.qmutator.com/docs/qflow): the visual automation engine
  and its node catalog.
- [Admin & MCP docs](https://api.qmutator.com/docs/admin): administration and
  MCP/agent setup.

## Site

- [Pricing](https://qmutator.com/pricing): all four modules (QLens, QSearch, QChat,
  QFlow) included with every account; unlimited catalog size; Azure-hosted.
- [Contact](https://qmutator.com/contact)

## Optional

- [Privacy](https://qmutator.com/privacy)
- [Terms](https://qmutator.com/terms)
- [Imprint](https://qmutator.com/impressum)