🚀 SWIRL Enterprise 3.8 Release Announcement

Team SWIRL is announcing the release of SWIRL AI Search 3.8, Enterprise Edition.

⭐ PLEASE STAR OUR REPO: https://github.com/swirlai/swirl-search
🌐 PLEASE VISIT OUR WEBSITE: https://www.swirlaiconnect.com/

Building on Community 3.8, SWIRL Enterprise 3.8 introduces SWIRL AI Co-Pilot for secure, intelligent assistance over your internal data (with no bulk copying or new vector DBs), streaming Chat/RAG LLM support, 2,000-token query limits for richer prompts and “query-by-example,” Reader LLM 2.0 with major speed and relevancy gains, a new Box connector with SSO, a new HTTP interface for Search & RAG, and the Galaxy Prompt Selector for fine-grained AI workflows.

🔹 SWIRL AI Co-Pilot
Turn any GAI/LLM into a secure assistant that answers with your internal repositories—on the user’s behalf—without moving data or deploying a separate vector database.

SWIRL AI Co-Pilot

🔹 New Box Connector (SSO)
Easily add Box as a source with single sign-on integration.

SWIRL AI Search RAG on Box.com

🔹 New HTTP Interface for Search & RAG
A streamlined API for search, re-ranking, document review, Q&A, and insight generation inside your apps and pipelines.

New features

  • AI Co-Pilot: Secure, enterprise-aware assistant; no bulk data movement or new vector DB required
  • Streaming Chat/RAG: Low-latency, incremental responses from LLMs
  • 2,000-Token Query Limit: Bigger prompts and dialog turns; enables query-by-example
  • Reader LLM 2.0: Faster embedding pipelines; better relevancy ranking
  • Box Connector (SSO): Federate Box content with enterprise auth
  • HTTP Interface for Search & RAG: Build search/RAG into existing solutions via API
  • Galaxy Prompt Selector: Fine-tune AI workflows with selectable prompts
  • PII Detection & Removal (MS Presidio): Optional redaction in pipelines

SWIRL AI Search with PII redaction by Microsoft Presidio

  • Updated RAG Citation Format: Clearer citation presentation in generated answers

SWIRL AI Search with new RAG citations

Improvements

  • Streaming Support for Chat/RAG LLMs
    Stream partial responses for faster, smoother interactions.

  • Query Limit Increased to 2,000
    Support more robust prompts, longer dialogues, and query-by-example use cases.

  • Reader LLM 2.0
    Up to 2× faster overall (~10× in Docker with OpenAI embeddings) plus improved relevance scoring to surface the best results.

  • Galaxy UI
    • Prompt selection integrated into workflows
    • Smoother streamed answer display and interactions
  • RAG / Search
    • Higher-quality citations in RAG results
    • Better result ranking via Reader LLM 2.0 improvements
  • Technical Improvements
    • Reader LLM throughput and latency gains (notably in Docker with OpenAI embeddings)
    • Expanded API surface via the new HTTP endpoints for Search & RAG

Upgrading

⚠️ Version 3.8 does not require database migration.

Documentation

📘 SWIRL’s documentation site has been updated reflecting the new features above.