Version 2.4.4 | Nov 27, 2025

Overview

ZBrain Builder 2.4.4 brings powerful enhancements to reasoning flexibility and enterprise control, as well as new workflow and orchestration capabilities. This release introduces new capabilities, including Knowledge Graph toggles for granular indexing and improved support for Microsoft Agent Framework in Agent Crew orchestration. Users can now convert Word and PowerPoint files to PDFs via Flow. These updates—alongside improvements in error detection, privacy settings, and dynamic scrolling—make ZBrain Builder more adaptive, composable, and enterprise-ready for real-world AI deployment.

ZBrain Builder 2.4.4 release overview

Component	Capability	What it delivers
Flows & Pieces	Visual alerts for missing inputs in Flow	Flags broken input links to streamline error detection and ensure seamless Flow continuity.
	File-to-PDF Conversion during knowledge ingestion	Enables seamless file conversion to PDF for consistent, ingestion-ready formatting.
Agents	Microsoft Agent Framework Support in Agent Crew Configuration	Expands orchestration flexibility by allowing agent crews to run on Microsoft’s framework alongside existing options
Apps	Comprehensive conversation logs for Flow and Agent Crew-powered Apps	Provides full traceability of app executions with step-level insights and metadata

New features

Flows

New Flow option: Convert files to PDF during knowledge ingestion

ZBrain Builder includes a new Flow action that allows users to automatically convert PowerPoint and Word documents into PDF format during data source configuration. This enhancement simplifies file normalization, ensuring consistent formatting and compatibility for downstream processing.

Navigation: ZBrain Builder → Knowledge → Create New → Data Source Configuration → Improve using Flow-> Convert document into PDF before text extraction

Key outcomes:

• Files are standardized to PDF format, reducing format-related ingestion issues across varied document types.

• Users no longer need to convert .pptx or .doc/.docx files externally before upload—ZBrain Builder handles the conversion inline.

• Ensures that content from common business file formats is easily usable for LLM-based enrichment or downstream AI workflows.

Agents

Support for Microsoft Agent Framework in Agent Crew configuration

ZBrain Builder supports the Microsoft Agent Framework as a selectable crew framework, enabling teams to build agent crews using Microsoft’s orchestration model alongside existing options, such as LangGraph, Google ADK and Semantic Kernel. This update enhances interoperability and provides flexibility in selecting agent coordination frameworks according to enterprise preferences.

Navigation: ZBrain Builder→ Agents → Create New → Create Agent Crew from Scratch / Update Existing

Key outcomes:

• Supports broader architectural alignment with Microsoft-based ecosystems.

Apps

Enhanced conversation log for Flow and Agent Crew-based apps

ZBrain Builder now provides full visibility into app executions by introducing comprehensive conversation logs for apps built using Flows or Agent Crews. This enhancement enables users to track, monitor, and analyze every query with detailed execution metadata, thereby enhancing transparency, auditability, and operational control.

Navigation: ZBrain Builder→ App → Select App → Query History-> Conversation Log → Run Details/ Crew Activity

Key outcomes

• Improves observability through step-by-step insights into how a Flow or Agent Crew processed a query, including status, timing, and token usage.

• Validates LLM behavior, traces execution history, and helps with transparent input-output logs.

Enhancements

Flows

Visual alerts for missing inputs in Flow

ZBrain Builder provides intuitive visual indicators when a step that supplies inputs to other workflow steps is deleted. This update enhances transparency and reduces rework by enabling users to quickly detect and resolve broken dependencies during Flow creation or modification.

Navigation: ZBrain Builder → Flows → Create New / Update Existing Flow

Key outcomes:

• Faster troubleshooting with clear exclamation icons and red highlights immediately shows where inputs are missing, reducing time spent identifying broken logic.

• Real-time feedback prevents incomplete steps from going unnoticed, improving flow reliability and business continuity.

• Users are guided through fixing impacted steps, minimizing the risk of failed executions or overlooked gaps.

• Visual cues persist until issues are resolved, ensuring that no broken step slips through during deployment.