Release Notes
All notable changes to the Agentic AI Manufacturing Framework at a glance.
[0.5.0-rev15] - 2026-04-02
Release summary
This release focuses on platform resilience, usability, and governance. Key improvements include critical bugfixes for session management and data retrieval, performance and stability hardening across core services, enhanced observability, and the introduction of Manifest Validation Policies - a new capability that gives operators fine-grained control over agent compliance.
What's New
Manifest Validation Policies
- The platform now supports configurable validation policies that automatically check agent and team manifests against operator-defined rules at deployment time. Policies use a flexible JSON Logic format, enabling operators to enforce security, compliance, and configuration standards - with violations surfaced directly in deployment responses and via a dedicated reporting endpoint.
Agent Operations
- Bugfix for session management when combined with external tool calling.
- Retrieval endpoint of the Semantic Query Engine now correctly handles multi-datasource configurations where one datasource was empty but existing, instead of returning a 404 when other datasources contained valid data.
Platform Stability and Performance
- Multiple performance and stability improvements across core platform services.
- Improvements and fixes for Data Ingestion Pipelines to increase reliability and throughput.
Security and Compliance
- General security fixes addressing identified vulnerabilities across platform components.
User Experience
- General UI and styling improvements for a more polished and consistent look and feel.
Monitoring and Observability
- Improved observability and metrics collection across platform components.
Operations and Onboarding
- Simplified Tenant Onboarding process to reduce setup effort for new customers.
What to Expect
- End users: A more stable and visually refined platform with reliable session handling and consistent data retrieval, even across complex multi-source configurations.
- Operators: New policy-based governance tools to enforce compliance standards on agent deployments, streamlined tenant onboarding, and deeper operational visibility through improved metrics.
- Developers: Enhanced observability for debugging, more resilient data ingestion pipelines, and a well-defined policy validation API for integrating compliance checks into automation workflows.
[0.4.0-rev19] - 2026-03-09
Release summary
This release delivers platform enhancements focused on customer integration, stability, and security. Improvements include expanded data connectivity options, enhanced agent operations, security vulnerability resolution, and improved monitoring and observability capabilities.
What's New
Data Connectivity
- Microsoft SQL Server and Access database integration for Business Analytics Platform.
- MS AccessDB Reader implementation for direct database connectivity.
- Data Ingestion Endpoint now adds Support for Custom Transformation automatically handled upon ingestion.
- Added new Custom Transformations to convert JSON to YAML or Markdown on Ingestion.
Agent Operations
- Stability Improvements for Tools used by Shopfloor Agent.
- User data deletion handling and improved startup resilience in Agent Runtime.
- Improvements to Shopfloor Agent timezone handling.
- Introduction of additional System Tests.
Security and Compliance
- Outdated libraries removed.
- Improvement to CI/CD Workflows.
- Updated Chat Interface Look and Feel.
- Enable Secret Management in Agent Runtime with tenant-specific Azure Key Vault.
- Added Endpoint to allow dynamic deletion of User Chat Sessions.
Monitoring and Observability
- Prometheus established as central metrics collector with OpenTelemetry integration.
- Component-level metrics enabled throughout platform.
- Network whitelisting for customer environment integration.
- OpenWebUI version 0.7.2 tested and deployed.
Framework and Architecture
- LLM Proxy service resilience improvements for Redis connection handling.
- Performance improvements for LLM Proxy component.
What to Expect
- End users: A stable platform with reliable agent interactions, expanded data connectivity and data transformations, and consistent performance across environments.
- Operators: Production-grade monitoring, unified dashboards for system health, reduced security vulnerabilities, and streamlined CI/CD processes.
- Developers: Standardized component architecture, flexible authentication options, secure dependencies, and improved observability for debugging.
[0.3.0-rev11] - 2026-01-30
Release summary
This release marks a solid step forward for the platform, focusing on reliability, security, and user experience. With several targeted improvements and upgrades, both end users and operators will notice smoother access, safer authentication, and better compliance across core services.
What’s New
Simpler, Stronger Access Control - Management of agents, teams, and data sources is now unified and streamlined. Authorization is handled consistently across the platform to simplify operations and reduce confusion.
Safer Authentication Everywhere - All user-facing interfaces and APIs now enforce secure login and permissions more strictly. Systematic testing ensures the right users have the right access—no more, no less. - Multi-factor authentication (MFA) is now available for extra security, especially for users without single sign-on.
Reliable Operations at Scale - Default limits help prevent one tenant’s heavy usage from affecting others, so performance remains stable even during peak times. - Key components for customer automation are now separated more cleanly from core platform services for added safety and resilience.
Smoother Integrations and Upgrades - The agentic backend is now compatible with the latest OpenAI chat completion standard thanks to the addition of a models endpoint. - Behind the scenes, core software components have been upgraded to more stable versions, closing off potential vulnerabilities and improving reliability.
New Data Sources and Smarter Agents - The platform now supports new types of data sources, including email integration, enabling agents to work with a broader range of business information. - Agent capabilities have been improved, making interactions more effective and flexible. - Our automatic testing framework for agents has been expanded, providing deeper measurement of agent quality and performance.
What to Expect
- End users: Expect secure, transparent login and consistently reliable access to all platform features.
- Operators: Benefit from easier role and permission management, plus simplified support and compliance checks.
- Developers: Enjoy improved integration options and a strong foundation for building and connecting new applications.
This release delivers foundational advancements while keeping everything our customers already rely on steady and dependable.
[0.0.4] - 2025-12-17
Release summary - current release
This release enhances system stability, strengthens security, and expands agent capabilities. Key improvements include robust infrastructure fixes, smarter operational tools, and optimized workflows for seamless deployment.
What's New
Agent Registry:
- Introduces an Agent Registry to create, modify, and compose agents and agent teams at runtime.
- Enabled dynamic orchestration and configuration without redeploys via a unified management API.
Elevated Operations & Integration:
- Reliable Build Workflow: Improved build pipelines to enhance the consistency and reliability of deployments.
- Terraform Deployment Improvements: Split infrastructure and application deployments to ensure faster and more reliable infrastructure provisioning.
- Updated Keycloak Integration: Aligned tenant onboarding scripts with new Keycloak requirements for streamlined authentication.
- Load Testing & CI/CD Enhancements: New load testing utility and streamlined build/deployment pipelines.
Advanced Agent Capabilities:
- Enhanced Shopfloor Agents: Smarter agent interactions with improved multi-turn conversations and more robust testing of Agent Behaviour.
- Optimized Data Ingestion: Introduced new, and improved existing Ingestion pipelines for smarter and reliable agent communication.
User Experience Improvements:
- Refined OWUI Design: Polished branding, and banner placements for a more user-friendly experience.
- Fixed Errors in Data Ingestion: Addressed content-encoding mismatches to reduce ingestion-related failures.
- Fixed Speech Configuration in OpenWebUI: Resolved an issue that led to errors when using Azure Speech TTS in OpenWebUI.
Improvements & Fixes
- Strengthened Security: Reinforced protection across the platform and resolved identified vulnerabilities.
- Enhanced Security Logging: New security logger across core services for better monitoring and compliance, with sensitive data redaction.
- Performance & Stability: Fixed agent runtime errors, and addressed API inconsistencies.
How This Affects You
- End Users: Enjoy improved AI interactions, a more secure platform, and smoother data processing.
- Operators: Experience streamlined workflows, simplified authentication management, and better operational reliability.
- Developers: Benefit from optimized build infrastructure, smarter CI/CD tools, and improved agent development frameworks.
Note
This release includes 12 additional fixes and updates not explicitly listed here.
[0.0.3] - 2025-09-26
Release summary — current release
We’re excited to ship multi‑tenant support and Single Sign‑On (SSO) in this release. These changes make it easier and safer to run multiple teams on the platform while letting users log in with the corporate identity provider.
What’s new
- Multi‑tenancy: Tenant separation supported end‑to‑end (data ingestion, semantic query engine, agent runtime). Organisations can run isolated data and configurations on our platform.
- Single Sign‑On (SSO): Login via Single KeyID is available across frontend and backend — unified corporate authentication.
- OpenWebUI update: New frontend version with improved startup behavior and easier admin onboarding.
- Prefect integration: Background jobs now run via an infra‑level Prefect + Azure Container Instances setup for more reliable task execution.
Improvements
- More consistent and robust data ingestion APIs.
- Performance and reliability work, plus refactorings to improve maintainability.
Notable fixes
- OAuth / token handling issues resolved (improved SSO and service token reliability).
- TenantID and SQE access bugs fixed so agent teams and ingestion work correctly in tenant mode.
- OpenAPI toolbox: improved discovery and parsing of API descriptions.
- Security patches applied across components.
How this affects you
- End users: sign in with corporate account; chat/frontend UX improved.
- Operators: follow updated onboarding docs for tenant setup and identity management configuration.