Version 2.3.3 | June 17, 2025

Overview

ZBrain Builder 2.3.3 release delivers smarter automation, granular resource control, richer monitoring, and clearer agent visibility, making it easier to build, operate, and govern enterprise-grade AI solutions. This version introduces new features and one improvement, making ZBrain Builder more reactive, resource-aware, and transparent.

Together, these updates make ZBrain Builder more event-driven, space-efficient, and transparent, empowering teams to deploy secure, scalable AI solutions with confidence.

New features

Flows & Pieces

Event-driven agents

Users can now add input triggers to any ZBrain agent, enabling it to run automatically whenever new data arrives, eliminating the need for manual polling or cron jobs.

Trigger-management highlights

• Queue-level trigger panel: Add, schedule, test, or delete triggers from the single ‘Create Queue’ step (Agents › Create New › Create Queue).

• Event-driven execution: Each trigger routes the incoming item (file, row, message) directly into the agent’s flow, starting the run instantly.

• Time-based trigger: The user can configure a time-based trigger by selecting a preferred interval in the additional settings.

• Safe rollback: The user can remove an existing trigger from the agent configuration at any time.

Event trigger types

• Amazon S3

• Discord

• Gmail

• Google Drive – File

• Google Drive – Folder

• Google Sheet

• Google Form

Key highlights

• Queue-level trigger management

• Event-driven execution

• Flexible scheduling and on-demand runs

• Safe rollback

SaaS

Storage management

This feature provides users with full visibility and control over how their default Knowledge Base storage quota is consumed through the Knowledge Storage dashboard, accessible from settings. Additionally, users can perform file-level management, enabling them to reclaim space.

The page displays the storage summary and limit, showing total usage (e.g., 3.2 GB Used) with a visual progress bar against the default storage limit. The ‘Buy Storage’ button lets users increase the storage limit.

The dashboard also displays the storage distribution, presenting a clear breakdown between Knowledge Base and Agent usage. Hovering over the Knowledge Base/Agents info icon displays “View files to free up storage space,” and clicking it opens the Manage Storage module.

The Manage Storage module lists every uploaded file, including the file name, associated KB, last accessed date, and size. Columns are sortable and multi-selectable. The user can select files, view the storage to be freed, and then click the ‘Delete’ button to confirm.

Key highlights

• Users can view up-to-date and detailed information about how their storage is allocated across different components, such as the Knowledge Base and Agent usage.

• Easy file cleanup instantly reclaims space, eliminating the need to purchase additional capacity.

• Scale up only when necessary, keeping usage optimized.

Evaluation framework

Configurable notifications through flows

Users can now create custom notification flows for any monitoring event by configuring event settings. These flows are triggered when the monitoring event meets the user's evaluation logic, i.e., the user-defined evaluation conditions and the specified frequency threshold.

• Users can configure notifications to be delivered via Email, Slack, Microsoft Teams, or any webhook-based integration. Users can use Email, Slack, Teams, or any webhook-based flow for notifications, which will be delivered on the same interval the user set for the evaluation (every 30 minutes, hourly, six-hourly, daily, weekly or monthly). A new ‘Send Notification’ toggle lets teams attach a notification flow from the Flow Library and test it instantly.

• Users cannot update the event settings until a valid notification flow passes the delivery test. If delivery fails, inline errors are displayed along with a link to edit the flow.

Key highlights

• Channel-flexible alerts are delivered directly to the appropriate teams through their preferred platforms, such as Email, Slack, Microsoft Teams, or any other webhook-supported channel.

• Rule-driven triggers activate only when exact metric thresholds are breached, thereby reducing false positives.

• Pre-defined interval settings ensure that alerts are perfectly in sync with the evaluation schedule, eliminating off-cycle noise.

• Test mode confirms delivery upfront, protecting production from silent failures.

• Inline validation and error prompts instantly catch misconfigurations and surface fixes directly on the page.

RAG

Incremental Graph-RAG growth

Users can now add a new chunk within a Graph RAG-enabled Knowledge Base by clicking the ‘+ Add Chunk’ button for the selected knowledge base.

• Key highlights

  • Incremental ingestion lets users append new content instantly, without the need to re-import the entire source.

  • Content is embedded and linked in the background, ensuring continuous access to Knowledge Base queries without interruption.

  • Real-time status cues switch from 'In Progress' to 'Active' the moment integration completes.

  • Append-only mode permits additions only, preventing accidental edits or deletions until a future update.

Improvements

Agents

Enhanced agent observability

This update brings enhanced visibility, classification, and interactivity to Crew Agents and the flow-based agents.

Key enhancements:

• Live agent activity log: Every Crew execution now streams a real-time log that shows the sequence of agent activation, the input each agent receives, its output, and any intermediate handoffs or decisions.

• Agent type classification: A new Type column in the Agents list makes it immediately clear whether an entry is an individual Agent (built with Flows) or a Crew (built with the Crew framework).

• Crew-structure clarity: Within the Crew editor, the currently selected node is outlined in blue, while unselected nodes retain a gray border, allowing for rapid visual focus as complex graphs grow.

Key enhancements:

• Greater observability exposes each step of an agent or Crew run through live activity streams.

• Clearer classification instantly distinguishes Flow-based standalone Agents from Crews with the new ‘Type’ label.

• An improved user experience streamlines build and operations through focused node highlighting and cleaner visuals.