A Magento-first AI content workflow with preview, review, and control

Rule creation

Use guided rule setup with templates, prompt presets, source field helpers, and model recommendations.

Product targeting logic

Build nested AND/OR conditions with grouped operators and insertable condition presets.

Two-stage filtering is preserved: Magento criteria first, exact local evaluation second.

Brand voice consistency

Create reusable brand voice profiles with tone, style rules, required phrases, and banned phrases.

Profiles can be scoped per account or store code.

Compliance guardrails

Apply lightweight text policy checks for banned phrases, required phrases, length, regex patterns, and superlatives.

No extra AI correction calls required.

Sample preview before run

Preview matching products and generated sample output before saving or executing broad runs.

• On-demand only
• Small sample caps
• No background preview jobs

Dry-run review and approval

Dry-run results can be reviewed item by item with old/new snapshots, then approved or rejected in bulk.

• Review status filters
• SKU search and reason insights
• Apply approved snapshots without regeneration

Magento publishing

Publish approved or direct apply output to Magento product attributes through existing API integration.

Run history and exports

Export run items as CSV by status or review state for QA, audit, and handoff workflows.

Skip optimization

Permanent source fingerprint skip avoids redundant AI generation when nothing changed.

Store connections and security

• Magento bearer and OAuth1 auth modes
• Encrypted credentials and AI keys
• Owner-scoped data access and CSRF/session protections
• Per-rule locking and retry behavior

API and CLI workflows

The platform remains API-first with web UI and CLI parity for scheduling and worker execution.

• /api/content-flow/rules/* and /api/content-flow/runs/*
• php cli.php content-flow:schedule
• php cli.php content-flow:work --loop