New Databricks industry report highlights the role of governance and evaluation in enterprise AI compliance and risk outcomes
Key takeaways:
- Enterprises with active AI governance deploy over 12× more AI projects into production than those without governance.
- Investment in AI governance and security tools grew 7× in nine months through October 2025, according to usage telemetry.
- Companies that use AI evaluation tools get nearly 6× more AI projects into production.
- Only 19% of organizations had broadly deployed AI agents despite widespread AI adoption.
- Multi-agent system use grew 327% over a four-month period, underscoring the rapid evolution of AI workloads.
Databricks released the 2026 State of AI Agents report, detailing enterprise trends in AI governance, evaluation, and compliance based on anonymized usage data from more than 20,000 organizations worldwide, including over 60% of the Fortune 500.
The report shows that unified AI governance and structured evaluation frameworks correlate with higher rates of AI project deployment and more consistent production outcomes.
Enterprises actively using governance tools deploy 12 times more AI projects into production relative to firms without such tools. Investment in governance and security technology increased sevenfold over a nine-month period through October 2025.
The findings also show that AI evaluation tools are linked to higher operational adoption. Organizations using evaluation tools were able to get nearly six times more AI projects into production.
The report places enterprise AI governance as a prerequisite for scaling autonomous agentic systems and underscores that integrating governance into AI workflows helps align development with business objectives and compliance requirements.
“As enterprises adopt agentic AI, old rules and processes are being turned on their heads,” the report states. “Unified governance helps enterprises ensure their AI development aligns with business goals, meets security and legal obligations, and accounts for regulatory risks.”
The report also highlights a 327% increase in multi-agent architectures over a four-month period. While overall AI adoption is widespread, the report states that only 19% of organizations deployed AI agents at scale as enterprises transition from pilot projects and chatbots to integrated systems.
The data was fielded from November 1, 2024, through October 31, 2025.