Skip to main content

The Knowledge Pattern

Actively Evolving

Agentic orchestration is one of the fastest-moving areas in enterprise software. Aprimo is actively investing in integrations, tooling, and guidance across orchestration platforms. The patterns documented here reflect our current best understanding — check back regularly as this space develops. Have a use case or question not covered here? Reach out to your Aprimo Customer Success Manager and we will work with you to understand your use case.

What This Pattern Does

The Knowledge Pattern makes Aprimo's content accessible to AI agents and assistants. Content is synced from Aprimo into your orchestration platform's knowledge index, where it can be retrieved and reasoned over. Users can ask questions in natural language and get answers grounded in your approved, governed Aprimo content.

important

This is a read-only pattern. If you need agents to take action in Aprimo, see the Invocation Pattern.


When to Use This Pattern

Use the Knowledge Pattern when the primary goal is surfacing information. For use cases that involve taking action, see the invocation pattern.

Example use cases:

  • "What are the approved key messages for the Q3 campaign?" → answered from campaign brief documents in Aprimo
  • "Do we have approved lifestyle photography for this product line?" → answered from record metadata
  • "What are our brand guidelines for social media?" → answered from brand guidance content

How It Works

Aprimo pushes content into your platform's knowledge index using the DAM Rules engine. DAM Rules fire on record lifecycle events — a status change, a file update, a field modification — and send the relevant record data to an external endpoint via an HTTP action. Your receiver processes the payload and forwards it to the platform's ingest API.

DAM Rule condition met

Aprimo HTTP action → your receiver endpoint

Receiver forwards record data to platform knowledge index

LLM reasons over indexed content → answers user's question

This push model keeps the index current without polling. When a record is approved in Aprimo, the update flows to the index immediately.

For more complex scenarios — conditional logic, filtering by classification, or transforming the payload before it leaves Aprimo — DAM Rules support expression building with logical operators so you can control exactly which records get pushed and when.

DAM Rules Help

See the DAM Rules documentation for configuration details.


What to Index

What your integration will need is dependent on your organization and workflows, but the following a guideline of what may provide value.

ContentWhy It's Valuable
Record metadata (title, description, tags)Makes records findable by natural language
Campaign briefs and documentsEnables Q&A over approved messaging
Product specs and brand guidelinesGrounds answers in authoritative content
Careful

Avoid indexing binary file content (images, videos) — index their metadata instead. Avoid indexing unapproved content unless you have a deliberate reason to do so.


Authentication

Aprimo supports OAuth 2.0 for all API access. For automated ingest pipelines, use the Client Credentials flow (service account). For scenarios where individual user permissions must be respected at query time, use Authorization Code with PKCE.

See the OAuth2 documentation for setup details.

Platform-specific considerations

Each orchestration platform handles data ingestion, authentication, and indexing differently. This article covers the Aprimo side of the integration. If you need help designing the connection between Aprimo and your specific platform, contact your Aprimo Customer Success Manager.

Permissiong Alignment Matters!

If a user shouldn't see a record in Aprimo, they shouldn't receive information from that record via the AI assistant either. Design your auth model accordingly.


Limitations

LimitationNotes
Read-onlyThis pattern surfaces content. For writes, use the Invocation Pattern.
Index latencyThere is a small delay between an event in Aprimo and the index being updated. For most use cases this is negligible.
Binary assetsImages and videos can't be reasoned over directly by most LLMs. Index their metadata, not their binary content.
Permission complexityIf different users should see different content, per-user OAuth at query time is more complex to implement but is the correct approach.