Product Changelog
Track what's new in RapidCore AI. We ship improvements weekly.
v1.4.0Latest
- NewMCP (Model Context Protocol) integration for tool-augmented retrieval
- NewVision PDF processing for scanned documents and image-heavy files
- NewGraph RAG support for relationship-aware retrieval across connected entities
- ImprovedRe-ranking accuracy +35% on technical documents
- ImprovedStreaming WebSocket performance under high concurrency
- FixedAccess control edge case with nested group permissions
v1.3.0
- NewSAP S/4HANA native connector for live data retrieval
- NewOracle Database direct integration via JDBC
- NewAdmin dashboard with usage analytics and query volume charts
- ImprovedChunking strategy is now fully configurable per data source
- ImprovedEmbedding model selection per data source (OpenAI, Cohere, local)
v1.2.0
- NewSelf-hosted LLM support — Llama, Mistral, and Qwen via Ollama or vLLM
- NewPluggable document splitters with configurable overlap and chunk size
- NewGDPR data deletion API for compliant right-to-erasure workflows
- ImprovedCold start performance 2× faster for first queries after idle periods
- ImprovedMulti-language retrieval quality for DE, PL, SV, and FR corpora
v1.1.0
- NewSharePoint and Confluence connectors with incremental sync
- NewRole-based access control at the retrieval layer — per-user document visibility
- NewCitation tracking with direct links to source sections
- FixedRate limiting incorrectly applied under high concurrency bursts
- FixedMemory leak in PDF processor for documents over 200 pages
v1.0.0Initial Release
- NewCore RAG pipeline with vector search and re-ranking
- NewOpenAI, Anthropic, and Mistral LLM integrations
- NewREST API with OpenAPI spec and API key authentication
- NewBasic document ingestion — PDF, DOCX, and TXT
- NewWeb dashboard for data source management and query testing