Enrichers, AI agents, and automation design

On this page

CluedIn can extend records with enrichers and AI-driven jobs, but those automation surfaces need architectural discipline. Otherwise they become another source of opaque changes instead of controlled improvement.

What you should get from this module

  • understand the documented built-in enrichers and AI-agent roles
  • choose enrichment and AI patterns deliberately
  • design automation with traceability and review in mind

builtin_agents_sp.png

Guided walkthrough

Begin with the enricher reference. The important lesson is not to memorize every package. It is to understand the categories of enrichment that CluedIn supports: company registries, maps, web sources, AI services, and custom or REST-based patterns. These enrichers can reshape records significantly by adding new data parts and external context.

Now return to the built-in AI agents page. The platform explicitly documents default roles:

  • Data Steward for fixing quality issues and looking for duplicates
  • Data Architect for creating rules

That means AI is being positioned as an accelerator for documented operational and architectural tasks, not as a magical replacement for design thinking.

Teach three architecture principles here:

  1. Scope before automation
    Always know which population the automation will touch.

  2. Review before trust
    Test jobs and inspect outcomes before broad enablement.

  3. Traceability after execution
    Make sure downstream teams can understand what changed, why it changed, and how to reverse course if needed.

Also connect enrichment to golden-record behavior. Every enricher or AI-driven correction changes the record’s shape over time. The architect should expect those changes to show up in record history, stewardship workflows, and sometimes downstream streams.

Role lens

Automation should reduce toil, not reduce explainability. The architect is responsible for making automation legible to stewards and safe for consumers.

Practice assignment

Pick one business domain and propose:

  • one suitable enricher category
  • one AI-assisted use case
  • the scope definition you would use
  • how you would test it
  • what success and failure would look like operationally

Exit criteria

  • The learner can describe when enrichers and AI are useful in CluedIn.
  • The learner can scope and test automation responsibly.
  • The learner understands the need for traceability after automated changes.

Suggested source material