Waiting for engine...
Skip to main content

Connecting a business glossary to your agent

Meta Hub is your semantic metadata layer in the Boomi Enterprise Platform, helping define organization-specific terms like "Active Customer". It stores business glossaries that provide a centralized, governed source of your organization's business terminology (definitions, acronyms, business rules, and relationships between key entities). Custom agents built in Agentstudio can leverage business glossaries to deliver accurate, trustworthy responses tailored to your organization.

You can attach one or more Meta Hub business glossaries to your custom agent built in Agentstudio so it can access your organization's terminology. Agents use this context at execution time to accurately interpret prompts and generate responses aligned with your organization's language.

Each glossary shows its endorsement status (Endorsed, Pending, or Deprecated) so you can make an informed decision about the context you are attaching.

Benefits​

Attaching a business glossary to your agent:

  • Improves response accuracy — Agents interpret organization-specific terms using governed definitions from Meta Hub.
  • Preserves context window space — Glossary content is retrieved on demand, not preloaded, and (with Extended Thinking on) retained across the session.
  • Reflects governed terminology — Glossaries are centrally managed and published in Meta Hub, ensuring your agent always uses current, approved definitions.
  • Supports informed decisions — Endorsement status (Endorsed, Pending, or Deprecated) helps you evaluate glossary quality before attaching it to an agent.

Prerequisites​

Before attaching a glossary, ensure:

  • You have an Agent Garden Developer role or Agent Garden Administrator role. Refer to Configuring developer privileges for privilege details.
  • You have access to Meta Hub glossaries.
    • View glossaries: META - User or View Association privileges.
    • Attach glossaries: META - Admin, META - Steward or Edit Association privilege.
    • Delete glossaries: META - Admin or Delete Association privilege.
  • You have an agent created in Agent Designer or are in the process of building an agent.

Important considerations​

  • Unpublished glossaries are not visible in Agent Designer.
  • You cannot create or edit glossaries within Agentstudio.
  • You can attach a maximum of 10 glossaries to an agent.
  • Glossaries are not included in the agent YAML export. After importing an agent that used a glossary, re-attach it manually. The exported PDF lists the glossary name for reference.
  • Before deleting a Meta Hub glossary, detach it from deployed agents.

Glossary data storage​

Glossary storage depends on the location of your Boomi Platform. Although your agent deploys to a specific runtime region, your glossary data stays in the location of your platform.

PlatformAvailable runtimesGlossary data location
USUS, GBR, Japan, AustraliaUS
European (GBR)GBR onlyGBR

Attaching a glossary to an agent task​

You can attach more than one glossary to a single task and attach glossaries to multiple tasks within the same agent.

  1. Navigate to Agent Garden > Agent Designer and open the agent you want to configure.
  2. Select the Tasks tab.
  3. Find the task you want to attach a glossary to, then select Glossaries.
  4. Select Attach glossary. A list of all published glossaries in your Meta Hub account appears.
  5. Select Attach next to the glossary you want to attach.
  6. Close the panel and select Save.

Deleted glossary behavior​

If a glossary is deleted in Meta Hub, it appears as Not available in Agent Designer. Agentstudio disables deployment until you detach it.

For deployed agents, the agent can no longer retrieve terms from that glossary, which may result in incomplete or inaccurate responses.

When does the Large Language Model (LLM) retrieve a glossary?​

Task instructions and glossary metadata indicate when a glossary is relevant. The LLM retrieves a glossary when a user asks a question whose answer involves a term defined in an attached glossary. You can also add specific task instructions to direct the LLM to bring relevant glossary data into the context during agent execution.

Can I review how the agent uses a glossary?​

Yes. The agent trace includes a new Retrieval step in addition to the existing Thinking and Actions steps (Refer to the Agent trace field reference for details). The Retrieval step that tells you the glossary retrieved, the glossary id, the glossary response, performance, and success status. If the glossary retrieval is successful, you'll see the glossary data retrieved.

Do glossaries affect the agent's context window?​

The glossary is not pre-loaded into the context window. The LLM uses the glossary's name and purpose to decide when to retrieve it, and only relevant portions are loaded on demand. With Extended Thinking enabled, the agent also retains retrieved glossary content for the rest of the session, so it doesn't need to retrieve the same definitions again on later turns. Refer to Agentstudio token limits for more information.

Next steps​

After attaching a glossary, test your agent to confirm it retrieves and applies your organization's terminology correctly.

On this Page