You'd be forgiven for thinking it's a little strange to bring AI-powered search to Google Workspace. After all, this is Google. The company whose name is synonymous with search. But while Google's public web search is peerless, Google Workspace is something else entirely. Enterprise data is messy, personal, and fragmented. It doesn't behave like the open web.

That's where SWIRL Enterprise 4.2 steps in. This release adds full support for Google Workspace, including Gmail, Calendar, Drive, and Contacts, all with OAuth2 authentication and optional OIDC auto-provisioning for seamless identity management.

Internal search is not Web search

The web is designed to be indexed. Enterprise data isn't. Your email attachments aren't linked. Your calendar invites aren't tagged. Google Workspace search is fast but it isn't built for cross-system context, RAG augmentation, or AI agents needing structured information across multiple silos. SWIRL is. When you ask a question like "what were the last three contract updates from our partners at BioTech Labs?", you don't want 50 results. You want an answer, with citations, permissions enforced, and context preserved.

New in 4.2: the SWIRL MCP server

SWIRL 4.2 ships with a new MCP server, deployable as a proxy, and fully equipped with authentication support. If you're building workflows in Crew.AI, N8N, or Microsoft Power Automate, this is the bridge you've been waiting for. The server exposes SearchProviders, AIProviders, and other key objects as resources, and serves both AI Search and AI Search Assistant APIs as tools.

Assistant updates: History, editing, and export

In this release, SWIRL AI Assistant gets an interface overhaul, adding ChatGPT-style chat history displayed on the left. Users can now edit, rename, and resume conversations without losing context or historical memory. Chat sessions can also be exported directly to PDF for sharing, printing, or secure storage.

Smarter search, smarter instructions

A new instruction field has been added to AI Search, allowing users or workflows to send direct instructions to the RAG LLM, giving more control and precision over how responses are generated. Whether you're looking for a concise summary, a bullet-pointed action plan, or a formal report-style answer, the instruction field lets you tailor the behavior of the agent without touching the prompt.

Unify without migrating

SWIRL doesn't force you to centralize your data. It unifies your knowledge without moving any of your information. You don't need a new data lake or to reformat everything into vectors. SWIRL searches data where it lives, using what you already have including Google Workspace, SharePoint, Confluence, Salesforce, and legacy databases.