Agentspace

name: agentspace description: Google Agentspace patterns for enterprise AI search and agents. Use when building unified search across 100+ enterprise connectors (SharePoint, Confluence, Jira, Salesforce, ServiceNow), identity-aware access, or custom Agentspace agents. tags: [gcp, agentspace, enterprise-search, connectors]

Google Agentspace

Build enterprise AI search and agent experiences across 100+ third-party data sources using Google Agentspace.

When to Use

• Building unified enterprise search across SharePoint, Confluence, Jira, Salesforce, ServiceNow, and more
• Creating custom AI agents within Agentspace for specific business domains (HR, IT, Finance)
• Implementing identity-aware search that respects source system ACLs
• Deploying NotebookLM Enterprise for document analysis at scale

Pre-Built Enterprise Connectors (100+)

Patterns

1. Create Data Store with Connector

from google.cloud import discoveryengine_v1 as discoveryengine

def create_confluence_data_store(
    project_id: str, location: str, data_store_id: str,
    confluence_url: str, space_keys: list[str],
):
    """Create a data store connected to Confluence."""
    client = discoveryengine.DataStoreServiceClient()

    data_store = discoveryengine.DataStore(
        display_name="Confluence Knowledge Base",
        industry_vertical=discoveryengine.IndustryVertical.GENERIC,
        solution_types=[discoveryengine.SolutionType.SOLUTION_TYPE_SEARCH],
        content_config=discoveryengine.DataStore.ContentConfig.CONTENT_REQUIRED,
    )

    operation = client.create_data_store(
        parent=f"projects/{project_id}/locations/{location}/collections/default_collection",
        data_store=data_store,
        data_store_id=data_store_id,
    )
    return operation.result()

2. Search Across Enterprise Sources

from google.cloud import discoveryengine_v1 as discoveryengine

def search_enterprise(project_id: str, location: str, engine_id: str, query: str) -> list[dict]:
    """Search across all connected enterprise sources."""
    client = discoveryengine.SearchServiceClient()

    request = discoveryengine.SearchRequest(
        serving_config=f"projects/{project_id}/locations/{location}/collections/default_collection/engines/{engine_id}/servingConfigs/default_search",
        query=query,
        page_size=10,
        content_search_spec=discoveryengine.SearchRequest.ContentSearchSpec(
            snippet_spec=discoveryengine.SearchRequest.ContentSearchSpec.SnippetSpec(
                return_snippet=True,
            ),
            summary_spec=discoveryengine.SearchRequest.ContentSearchSpec.SummarySpec(
                summary_result_count=5,
                include_citations=True,
                model_spec=discoveryengine.SearchRequest.ContentSearchSpec.SummarySpec.ModelSpec(
                    version="gemini-1.5-flash-002/answer_gen/v2",
                ),
            ),
        ),
    )

    response = client.search(request)
    results = []
    for result in response.results:
        doc = result.document
        results.append({
            "title": doc.derived_struct_data.get("title", ""),
            "snippet": doc.derived_struct_data.get("snippets", [{}])[0].get("snippet", ""),
            "link": doc.derived_struct_data.get("link", ""),
            "source": doc.derived_struct_data.get("source_type", ""),
        })
    return results

3. Custom Agentspace Agent

Agentspace allows creating domain-specific agents that combine search + actions:

IT Help Desk Agent:
  Data Sources: ServiceNow KB, Confluence IT docs, Jira known issues
  Actions: Create ServiceNow ticket, Escalate to on-call
  Identity: Uses Google Workspace SSO, respects ServiceNow ACLs

HR Benefits Agent:
  Data Sources: SharePoint HR policies, Workday benefits docs
  Actions: Open HR case, Schedule benefits consultation
  Identity: Uses Google Workspace SSO, scoped to HR data only

4. Identity-Aware Access Control

Agentspace automatically respects source system permissions:

• Google Workspace: Uses Google Workspace ACLs (Drive sharing, Gmail access)
• SharePoint: Maps Azure AD groups to search results filtering
• Confluence: Respects Confluence space and page permissions
• Salesforce: Uses Salesforce profile and sharing rules

No additional configuration needed -- the connector framework handles ACL passthrough.

Key Capabilities

• NotebookLM Enterprise: Embedded document analysis with Audio Overviews
• Multimodal Search: Text, image, and document search across all connected sources
• Actions: Agents can create Jira tickets, update ServiceNow records, send Slack messages
• Grounding: All responses cite source documents with links
• Admin Console: Central management of connectors, agents, and access policies

Anti-Patterns

• Connecting all data sources without access review -- audit ACLs before enabling connectors
• Using Agentspace for real-time transactional data -- it's optimized for search, not OLTP
• Skipping connector sync schedule configuration -- stale data leads to poor answers
• Not configuring data exclusions -- exclude sensitive folders/spaces from indexing

References

• Google Agentspace
• Agentspace Connectors
• Vertex AI Search