Trust3 AI hooks into Databricks Agent Bricks for enterprise AI governance

Trust3 AI said June 17, 2026, that it has integrated with Databricks Agent Bricks to give enterprises a separate trust layer for monitoring, governance and control across AI agents running in multiple systems. The move targets a growing problem for businesses deploying autonomous agents at production scale: shadow agents, weak visibility and rising security risk.

Why it matters: - Enterprises are moving AI agents from pilots into production, but those agents often operate across multiple models, tools, data systems and orchestration frameworks. - That creates blind spots for governance, security and compliance teams. - Trust3 AI is positioning its platform as an independent control layer that follows agents wherever they run. - The company says the goal is to help organizations deploy autonomous agents responsibly at production scale.

What happened: - Trust3 AI announced an integration with Databricks Agent Bricks on June 17, 2026. - The integration adds unified visibility, governance and control over AI agents, regardless of where those agents run. - The company also said it plans to integrate with Databricks’ open-source project OmniAgent through a custom policy engine plugin. - Trust3 AI included a demo link in its announcement: Book a demo.

The details: - Databricks Agent Bricks is a governed enterprise platform for building and scaling AI agents. - The platform supports model choice, MCP-based integrations, unified governance and end-to-end monitoring. - Trust3 AI extends that control plane into environments where agents do not stay inside one system. - The integration is designed to provide cross-platform trust enforcement, runtime observability and token usage monitoring. - Trust3 AI said the OmniAgent plugin would let organizations apply tailored governance controls across custom agent workflows. - The Trust3 AI Trust Score assigns each governed agent a number from 1.0 to 10.0. - Score bands map to trust and risk levels, with 7.75 and above classed as High trust with Limited risk. - Scores below 4.15 indicate Critical trust gaps and High operational risk. - Scores fall when policy violations remain unresolved, agents lack assigned ownership, documentation is incomplete or unapproved models are detected. - Scores rise as teams remediate violations, assign owners and complete required fields. - Trust3 AI said token usage should be treated as a trust signal, not just a cost metric. - Sudden token spikes can point to runaway agents, prompt abuse, misconfigured tools or unintended data access. - The platform surfaces those patterns in real time. - Trust3 AI said the system helps teams catch issues before they become operational failures or security incidents.

Between the lines: - The integration reflects a broader shift in enterprise AI from model oversight to agent oversight. - Traditional governance tools were built for more controlled systems and can miss unsanctioned agent activity. - Trust3 AI is arguing that trust controls need to be platform-agnostic because agents spread across stacks. - The customer example in the release suggests the company wants to show fast detection of shadow agents as a concrete use case. - One large oil and gas company reportedly found shadow agents running outside approved processes within less than five minutes of deploying the Trust3 AI collector. - The same customer-backed account said the company quickly shut down agents that were not supposed to be active.

What's next: - Trust3 AI plans to bring its policy engine to OmniAgent. - The company also plans to join the Databricks Built On program. - Trust3 AI said enterprises can use the integration to move faster while keeping governance, security and compliance controls in place. - The company will continue marketing the platform as an independent trust layer for AI systems across frameworks, clouds and data sources.

The bottom line: - Trust3 AI is betting that the next enterprise AI battleground is not just model quality, but continuous control over agents that move across systems and create new operational risk.