12/17/2024 - v2.2.1
Tailwinds 2.2.1 Release Notes
Expanding Our AI Model Ecosystem
We're excited to announce a major expansion of Tailwinds' AI capabilities with support for several cutting-edge language models across multiple platforms:
Amazon Bedrock Nova Series: Full integration with Amazon's latest Nova models (Pro, Lite, and Micro), providing flexible options from lightweight tasks to complex enterprise applications
IBM watsonx.ai & Granite 3: Enterprise-grade AI with IBM's powerful Granite 3 models, delivering superior performance for business-critical workflows
Google Gemini Flash 2.0: Lightning-fast inference and enhanced multi-modal capabilities from Google's latest innovation
We're also happy to announce a significant UI revamp with our new branding, interactive elements and more!
With these new additions, Tailwinds now offers unprecedented choice in AI models while maintaining our commitment to simplicity. Switch between models with a few clicks, compare performance, and find the perfect balance of speed, cost, and capability for your specific needs.
Let's dig in!
New Model Support
Nova Pro: Premium performance for complex enterprise applications
Nova Lite: Optimized for efficient, cost-effective processing
Nova Micro: Streamlined for quick, lightweight tasks
Full integration with existing AWS infrastructure
Enhanced performance metrics and cost management
Optimized for enterprise-scale deployments
Full support for IBM watsonx.ai platform
Granite 3 series models now available
Enhanced natural language understanding
Improved contextual awareness
Superior performance for enterprise workloads
Seamless integration with existing IBM Cloud infrastructure
Advanced prompt engineering capabilities
Enterprise-grade security and compliance features
New Gemini Flash 2.0 Experimental integration
Significant performance improvements over previous versions
Enhanced speed and efficiency for real-time applications
Improved multi-modal processing capabilities
Better context handling and response accuracy
Multi-Model Ecosystem
Groq Integration
Added Groq Llama3.2 support with optimized inference
Enhanced performance for large-scale language tasks
Configurable parameters for speed/quality balance
Gemini Model Suite
Integrated gemini-1.5-flash-8b for improved performance
Enhanced multi-modal capabilities for rich content processing
Extended AI Provider Support
New Alibaba API credential and ChatAlibabaTongyi node
Cerebras integration for specialized AI workloads
Jina AI Embedding support for enhanced vector operations
Enhanced Metadata Systems
Extract Metadata Retriever
Advanced document property extraction
Configurable metadata schemas
Automated tagging and categorization
Vector Store Improvements
MongoDB Enhancements
Enabled filters in MongoDB Vector Search Retriever
Improved query performance and accuracy
Enhanced scaling capabilities
PostgreSQL Integration
New Postgres agent memory implementation
PGVector Driver options with TypeORM compatibility
Resolved null character handling issues
Couchbase Support
Added Couchbase vector store integration
Optimized for enterprise-scale deployments
Enhanced data persistence and retrieval
3. Security and Enterprise Features
Advanced Security Framework
Configuration Management
New Security tab in Flow Configuration
When using the API, you must now explicitely enable the ability to Override a configuration (previously this was on by default)
Monitoring and Analytics
Metrics Collection
Prometheus integration for comprehensive monitoring
Open Telemetry support for distributed tracing
Custom analytics override configurations
User Experience and Interface Improvements
Chat and Interaction
Enhanced Chat Capabilities
Improved message history management
Follow-up Prompts feature for natural conversations
Interface Optimization
Workflow Management
Autofocus improvements for chat flow titles
Keyboard shortcuts for common actions
Visual feedback for node connections
Document Handling
File Processing
Improved document upload reliability
Enhanced file extension handling
Better support for special characters in filenames
UI/UX Improvements
Interface Reliability
Animations to help denote how nodes are interconnected
New homepage with examples workflows and guides
Fixed boolean handling in UI components
Improved markdown rendering
Enhanced search functionality
Better keyboard shortcut implementation
Breaking Changes and Migration Notes
Configuration Changes
Override Config is now disabled by default for API and Embed
Security tab must be used to enable specific override properties
Known Issues:
Amazon Nova
Due to an issue with the underlying LangChain framework, Nova models currently do not Tool call within an Agentflow (but will Tool call in a Chatflow).
Watsonx.AI
Due to an issue with the underlying LangChain framework, Granite models currently do not Tool call within an Agentflow (but will Tool call in a Chatflow).
Gemini Flash 2.0 (GF2E) Experiment
Due to an issue with the underlying LangChain framework, the GF2E model currently does not Tool call within an Agentflow (but will Tool call in a Chatflow).
Last updated
