Overview:
SWIRL is an open-source AI search platform that enables teams to perform ChatGPT-level search across their private data sources without moving data to the cloud. It connects to tools like SharePoint, Confluence, Drive, GitHub, and Jira, allowing users to ask questions and receive answers with source links. Designed for organizations seeking to deploy secure, self-hosted search, SWIRL respects existing permissions and eliminates the need for vector databases, ETL pipelines, or complex infrastructure setup.
Core Features:
Microsoft 365 Integration: Includes OAUTH2 support for direct connection to OneDrive and other Microsoft 365 services.
Search Federation: Performs synchronous or asynchronous search across multiple connected data sources, with results unified and re-ranked.
Real-Time Processor Pipelines: Adapts and transforms queries, responses, and results in real-time through configurable processor stages.
Built-in RAG (Retrieval Augmented Generation): Supports real-time RAG via OpenAI API, enabling AI-generated answers from enterprise data.
Duplicate Detection & Re-ranking: Detects duplicates via field comparison or Cosine Similarity, and re-ranks unified results using spaCy-based semantic similarity.
Extensible Connector & Mixer Architecture: Provides 100+ enterprise connectors and the ability to create custom connectors and result mixers.
Use Cases:
Knowledge base search: Teams connect SharePoint, Confluence, and Google Drive to get instant answers with source links while keeping sensitive data secure.
Customer support assistant: Support teams search across documentation and tickets, then draft responses using their own content to maintain consistency.
Developer assistant: Engineers search GitHub, Jira, and documentation simultaneously to find code examples and solutions, speeding up development workflows.
Unified enterprise search: Organizations provide a single search interface across all internal tools, with results that respect existing user permissions.
Why It Matters:
SWIRL offers a self-hosted alternative to cloud-based enterprise search tools, allowing organizations to keep data in place and on their own infrastructure. It eliminates the need for data duplication, vector databases, or complex ETL pipelines. With 100+ connectors, optional RAG via OpenAI, and permission-respecting search, SWIRL provides a practical path to AI-assisted search for teams that prioritize data control and rapid deployment.
