Prebuilt agent workflows
CataZenta ships ten ready-made workflows in the Agent Control Center. Each workflow runs a sequence of specialist agents on the catalog scope you choose — one product, a category, a product family, or a broader catalog slice.
Every proposed catalog change still goes through the approval queue before it is applied. Workflows can pause when a step needs your review, then continue after you approve.
Prebuilt workflows live in PIM under Agent Control Center → Workflows. They orchestrate in-platform specialist agents.
AI workflow examples are business playbooks (Amazon-ready content, publish gates, multilingual catalogs) that may combine PIM, APIs, and optional MCP — they are not the same as these ACC templates.
PIM workflows & tasks are human approval stages (legal, brand, ops) — complementary to agent proposals.
How to start a prebuilt workflow
- In PIM, open Agent Control Center from the sidebar.
- Go to Workflows → New workflow.
- Choose a prebuilt workflow (not “custom” with a free-text goal).
- Pick scope — single product, category, product family, or catalog-wide where supported.
- Name the run (a suggested name is filled in from the template).
- Select Create, then Start.
- Watch the timeline on the workflow detail page; open Approvals when status is Awaiting approval.
- Use Orchestration (live board) to see active runs across the tenant.
For a single test recipient or SKU, you can also run one specialist agent manually from Agents — workflows are for multi-step operations at scale.
Choose the right workflow
| Your goal | Start with this prebuilt workflow |
|---|---|
| Fill gaps, improve SEO, check compliance before publish | Catalog enrichment pipeline |
| Go live on marketplaces with data + compliance + syndication | Marketplace launch pack |
| EU sustainability / Digital Product Passport programs | Sustainability & DPP |
| Fix duplicates, taxonomy, and variant structure | Taxonomy & duplicate cleanup |
| Deep compliance only when enrichment finds critical issues | Critical compliance escalation |
| Recover after a failed enrichment pass | Failed run recovery |
| Multi-locale content and compliance | Localization & global readiness |
| Image-heavy catalogs — assets before copy and SEO | Media-first enrichment |
| New supplier or ERP import — onboard to publish-ready | Import onboarding pipeline |
| Fast SEO pass without compliance gating | SEO content sprint |
Workflow catalog
Catalog enrichment pipeline
Best for: Day-to-day catalog quality on products, categories, or families.
What it does: Runs three agents in order — fill missing attributes and content, optimize search metadata, then validate compliance.
| Step | Specialist agent |
|---|---|
| 1 | Product enrichment |
| 2 | SEO optimization |
| 3 | Compliance |
Typical scope: One product, a category, or a product family.
Marketplace launch pack
Best for: Preparing a slice of the catalog for marketplace syndication.
What it does: Enriches data, validates compliance, then runs syndication readiness and SEO optimization in parallel.
| Step | Specialist agent |
|---|---|
| 1 | Product enrichment |
| 2 | Compliance |
| 3a | Syndication readiness (parallel with 3b) |
| 3b | SEO optimization (parallel with 3a) |
Typical scope: One product, a category, or a product family.
Sustainability & DPP
Best for: Regulated sustainability and Digital Product Passport initiatives.
What it does: Base enrichment, DPP metadata generation, then compliance review.
| Step | Specialist agent |
|---|---|
| 1 | Product enrichment |
| 2 | DPP readiness |
| 3 | Compliance |
Typical scope: One product, a category, or a product family.
Taxonomy & duplicate cleanup
Best for: Structural catalog hygiene after imports or mergers.
What it does: Finds duplicates first, then runs taxonomy normalization and variant grouping in parallel.
| Step | Specialist agent |
|---|---|
| 1 | Duplicate detection |
| 2a | Taxonomy (parallel with 2b) |
| 2b | Variant grouping (parallel with 2a) |
Typical scope: Category, product family, or catalog-wide. Prefer a bounded category or family over the entire catalog when possible.
Critical compliance escalation
Best for: Targeted compliance depth when enrichment surfaces critical findings — not for routine bulk passes.
What it does: Always runs enrichment; runs an extra compliance pass only if the enrichment step reports critical insights.
| Step | Specialist agent | Notes |
|---|---|---|
| 1 | Product enrichment | Always runs |
| 2 | Compliance | Runs only when step 1 has critical findings |
Typical scope: One product, a category, or a product family — avoid catalog-wide runs for this template.
Failed run recovery
Best for: Recovering SKUs where enrichment did not complete successfully.
What it does: Retries enrichment; if enrichment fails, follows with asset intelligence (images and media metadata).
| Step | Specialist agent | Notes |
|---|---|---|
| 1 | Product enrichment | Always runs |
| 2 | Asset intelligence | Runs only if step 1 fails |
Typical scope: One product or a product family.
Localization & global readiness
Best for: Multi-locale catalogs before regional publish.
What it does: Base enrichment, translation gap analysis, then compliance.
| Step | Specialist agent |
|---|---|
| 1 | Product enrichment |
| 2 | Translation |
| 3 | Compliance |
Typical scope: One product, a category, or a product family.
Related playbook → Multilingual catalog
Media-first enrichment
Best for: Fashion, home, and other catalogs where asset quality drives merchandising.
What it does: Asset analysis first, then content enrichment and SEO.
| Step | Specialist agent |
|---|---|
| 1 | Asset intelligence |
| 2 | Product enrichment |
| 3 | SEO optimization |
Typical scope: One product or a product family.
Import onboarding pipeline
Best for: New feeds from ERP, PLM, or spreadsheets.
What it does: Duplicate check, enrichment, taxonomy fix-up, then compliance.
| Step | Specialist agent |
|---|---|
| 1 | Duplicate detection |
| 2 | Product enrichment |
| 3 | Taxonomy |
| 4 | Compliance |
Typical scope: Category, product family, or catalog-wide — start with the imported slice, not the whole tenant on day one.
SEO content sprint
Best for: Quick discoverability improvements without a compliance gate in the same run.
What it does: Enrichment then SEO only (two steps).
| Step | Specialist agent |
|---|---|
| 1 | Product enrichment |
| 2 | SEO optimization |
Typical scope: One product, a category, or a product family.
Related playbook → Amazon-ready content
Custom workflows vs prebuilt
| Prebuilt workflow | Custom workflow | |
|---|---|---|
| Setup | Pick a template; steps are ready immediately | Describe a goal in your own words; CataZenta plans the steps |
| Best for | Repeatable, proven catalog operations | One-off or experimental multi-agent plans |
| Wait time | No planning delay — review steps and start | Short planning step while the plan is generated |
| Steps | Fixed, tested sequences | Flexible; may include conditional branches |
Use prebuilt workflows for production catalog operations; use custom when you need a bespoke multi-agent plan.
Scope and batch size
| Scope | What it means |
|---|---|
| Single product | One SKU — ideal for pilots and high-value items |
| Category | All active products in a category — common for seasonal updates |
| Product family | Products sharing a family template — good for attribute consistency |
| Catalog | Broader slice — use for taxonomy/import templates; start smaller when learning the ACC |
Large scopes run one agent job per product per step. Review Activity and Approvals regularly; approve in batches when proposals look correct.
After the workflow runs
- Insights — findings with severity appear under Insights and roll into Catalog quality.
- Approvals — proposed attribute and content changes wait in the approval queue.
- Syndication — after approvals, publish channel-ready SKUs with syndication and export jobs.
Run Catalog enrichment pipeline on one incomplete product, approve proposals, then repeat on a category before marketplace or import templates.
Related documentation
- Agent Control Center — pages, permissions, governed loop
- Specialist agents — what each agent does
- Approvals — review queue vs PIM workflows
- AI workflow examples — business playbooks (different from ACC prebuilt workflows)
- MCP agent workflows — patterns for external AI clients (Claude, Cursor, etc.)