🎉 New Features
Schemas
AI-Assisted Schema Generation from Prompt
Automatically create your Schemas from natural language prompts. Schema creation now supports prompt-driven generation and inline import flows. Generated outputs are parsed from markdown/JSON, converted to Ligantic Schema config, and pre-populated into Schema creation with suggested naming/display metadata when available.
This feature is currently feature-flagged in Beta testing. Please reach out to support@ligantic.com if you want to join the list.
List Length Validation Rules
List fields now support minLength and maxLength validation so you can constrain collection size directly at the Schema level. You can also combine the validation rules to ensure you always get at least 1 but no more than 5 list entries (for example):
Geography Reference Support for Map-Oriented Fields
Schema and map-related editing/rendering now support expanded geography reference behaviour, including generated geography assets and registry-backed map packs for richer region-level mapping workflows.
Flows
AI Router Node: Smart Logic for Dynamic Workflows
The AI Router Node is a new addition to the Flow designer and gives you the power to add conditional logic powered by LLM decision-making to your Flows. Unlike standard If or Switch control nodes (both still available), the AI Router evaluates context to select the most appropriate execution path and can output route-specific structured data.
AI Router Node: Core Functionality
Intelligent Route Selection: Uses LLM reasoning to analyse inputs and determine the correct path based on your defined options.
Flexible Model Configuration: You can select a specific model at design time or use dynamic inputs to pass a Model ID during execution. This enables token optimisation by routing simple tasks to smaller models and complex tasks to larger models.
Instructional Framework: Provide optional system prompts or instructions to give the model specific criteria for its routing decisions.
Variable Integration: The node supports dynamic input variables using any properties available within the Ligantic Platform.
Structured Outputs & Control Flow
Each route defined within the node can be configured with its own Structured Output. By enabling this, you can define a specific schema (e.g., an object containing name, email, or intent) that is only populated and passed forward if that specific route is selected.
This ensures that subsequent nodes in your workflow receive clean, predictable data structures tailored to the chosen path.
Transparency and Debugging: To help manage and audit automated decisions, the AI Router includes:
Model Reasoning: An option to capture the internal logic used by the model to reach its decision.
Selected Option Details: Captures metadata about the chosen route for easier logging and flow monitoring.
Common Use Cases
Inbound Triage & Management: Automatically distinguish between a support request, a sales lead, or a general query, extracting relevant contact details only when a form submission is detected.
Cost Management: Evaluate the complexity of a user query and route it to either a high-performance model for reasoning or a cost-efficient model for simple retrieval.
Intent-Based Automation: Trigger different sub-flows based on the detected intent of a message, providing a more "agentic" approach while maintaining a structured, predictable workflow.
Video Audio Extract Node
New media:video:audio:extract Node was added to extract audio tracks from video inputs, with configurable output format and audio quality options.
Expression format() Optional Timezone
Data expression formatting now supports an optional timezone argument, enabling deterministic date/time output formatting across locales.
Dynamic/Expanded AI Model Coverage
AI flow model selection and model lists were expanded to include new OpenAI/Claude options and to continue handling compatibility for model evolution.
Experiences
URL Parameter Precedence for Entity Loading
Experience entity loading now supports URL parameter precedence over entity ID loading, making deep-linking and query-driven experiences more predictable.
Search Value Patching into Entity Context
Search values are now patched into the entity context instead of overwriting it, reducing state loss during query-driven rendering.
Spaces
Session Inactivity and Lifetime Enforcement
Authentication/session stack now supports configurable idle and absolute session expiration behaviour.
Change Password in Account Settings
Users can update passwords from settings, with validation handling integrated into the server-side policy flow.
Post-Login "Stay Signed In" Prompt Placement
The "Stay signed in" interaction was moved to a post-login prompt to clarify session intent and improve the sign-in flow.
🐛 Fixes
Schemas
Experiences
Fixed input block path-config error presentation.
Prevented unnecessary array identity regeneration on each render.
Fixed editor focus-loss and drag/scroll issues in complex editors.
Fixed no-refresh/refresh edge cases in auth-related experience flows.
Flows
Fixed Google AI chat-completion in-memory file handling.
Improved execution validation compliance for foreach behaviour.
Fixed mapping/comparison issues in flow processing paths.
Spaces
