Team SWIRL is announcing the release of SWIRL Enterprise 5.0 - the biggest release in SWIRL's history. SWIRL 5 adds a knowledge authority layer on top of federated search: it doesn't just find your organization's information, it identifies which version your organization actually trusts, and serves that answer to your people and your agents.
Semantic Cache and Canonical Version Finding
The new Semantic Cache (separately licensed) caches results from your connected sources, clusters document versions across systems, and elects the canonical version - the one your organization should be using. Curators can pin an organizationally endorsed answer for a query, and cached results carry freshness verification so stale content is flagged before it misleads. RAG answers are pinned-version aware.
Three-Pass Relevancy Model
SWIRL 5 ranks results with three passes: term and phrase match scoring, embedding-based semantic similarity, and cross-encoder re-ranking of the top results (enabled by an AI Provider with the reranker role). The result is relevancy that holds up across 150+ heterogeneous sources - without moving your data into a central index.
Streaming Search, Fetch More, and the Topic Dashboard
- Async, streaming search: results stream in as sources respond, with live per-stage progress, token-by-token AI Insight streaming, and a Stop button for both search and AI Insight generation
- Fetch more results: retrieve additional results on demand from every source that has more, without re-running the search
- Topic Dashboard: a personal home page of search topics with one-click folder caching and a configuration-driven Recent Files section for any connected OAuth2 source
- Duplicate handling: near-duplicate results are grouped rather than dropped, with a Show/Hide duplicates toggle
- Result provenance: every result carries its source, the query as sent, the fetch timestamp, and the response time
Multi-Model AI and the Assistant
AI Providers now cover all major LLM platforms via LiteLLM and direct connections, with per-provider models and prompts, per-family prompt overrides, per-model RAG context budgets, streaming, and structured outputs. The AI Search Assistant adds asynchronous chat with live progress, Markdown table rendering, answer feedback, and RAG answer persistence - AI Insights survive page reloads and are recorded for review.
A First-Class MCP Server
SWIRL 5 ships a built-in MCP server with stdio and streamable HTTP transports, 8 tools, and read-only deployment resources - giving Claude, Copilot, and your own agents permission-checked access to enterprise knowledge, the safe way.
Workspaces, Governance, and the Business Console
- Workspaces and members with owner, partner, editor, and viewer roles and invites
- Keycloak (OIDC) login, with configurable brokered access to Microsoft 365 sources
- OAuth refresh-token handling: connections to sources stay live across sessions
- Business Console: redesigned admin home with 30-day search and provider charts, Activity Analytics, a live Log Viewer, connector failure logging, create-by-pasting-JSON for any object, and white-label branding
- Query transforms: configuration-driven query rewriting per SearchProvider, with CSV bulk-load of rewrite and synonym rules
Platform
Python 3.13, updated Apache Tika with OCR enabled by default, in-process passage detection for RAG, and Docker-first deployment via Docker Compose with pinned sidecars for Qdrant, Redis, Tika, and local Ollama models.
Full release notes are available in the SWIRL documentation. Want to see it on your own stack? Talk to us.