Automation

AI Agents & EntiHubCLI

AI agents can accelerate MDM operations, but only when they run inside a clear governance and deployment model. EntiHubCLI gives that model.

What AI agents should do in EntiHub

Translate business intent into draft YAML entity definitions.
Generate repeatable CLI commands and CI/CD steps for deployment.
Prepare safe imports and validation checks before any publish action.
Assist stewards with change summaries, impact explanations, and rollback-ready plans.

Stage	Agent responsibility	Human responsibility
Design	Draft entity YAML and validation proposals	Approve business semantics and ownership model
Delivery	Prepare CLI/pipeline steps and dry runs	Review pull request and release plan
Governance	Provide impact notes and anomaly hints	Approve/reject sensitive changes
Operations	Generate runbooks and repetitive command chains	Own production decision rights and exceptions

PR-first policy: all model changes go through pull request review.
Environment separation: separate credentials and keys for dev, test, and production.
Least privilege: role-scoped API keys and minimal command scope.
Approval gates: keep mandatory approvals for sensitive entities.
Audit visibility: every important action remains traceable.

A typical agent-assisted flow from model to first data load:

entihub entity create --from-yaml customer.yaml
entihub entity deploy customer
entihub data import customer --file seed.csv

In production, run through CI/CD and protected approvals, not as direct ad-hoc changes.

New entity accelerator: "Create supplier entity with tax and bank validation" -> draft YAML + deployment steps.
Bulk onboarding assistant: map source columns to entity fields and generate import prep checks.
Change risk assistant: summarize what a schema or rule change can impact before release.