#USA #AI Governance #Vienna #Redwood Software #Agent-Based AI

The Rising Costs of Uncontrolled AI Agents

Recent research from Redwood Software, a leader in orchestration platforms for autonomous enterprises, highlights significant challenges today’s businesses face regarding ungoverned AI agents. The study, titled 'Enterprise benchmarks in AI and agentic orchestration 2026,' surveyed 300 high-ranking executives specializing in technology, risk management, and AI practices. Findings from the report indicate that a staggering 77% of companies have experienced adverse effects due to lack of control over AI agents, leading to unanticipated costs and operational inefficiencies.

Key Findings

The data collected through an independent panel by Leger Opinion emphasizes that while organizations are eager to expand their use of AI, the integration and governance necessary for safe and sustainable operations lag behind. Frequent negative outcomes include unexpected cloud expenses, business disruptions, and compliance-related issues.

1. AI Expansion vs. Governance: While 89% of respondents observed an uptake in agent-based AI usage over the past year, a mere 42% reported possessing robust systems to connect AI agents with their existing platforms. This underscores a fundamental disconnect between the aggressive deployment of AI and the necessary governance structures that should accompany such decisions.

2. Fragmented Responsibility: The responsibility for managing agent-based AI does not lie with a single entity. Different functions—IT operations, specialized AI and automation teams, and software developers—share the responsibility. This division often results in inconsistencies in integration quality, transparency, and governance compliance.

3. Risk Factors: The findings reveal that business risks are already present in organizations deploying AI systems without effective oversight, leading to governance strategies that failed to keep pace with implementation initiatives. Executives frequently extend autonomy to AI systems without comprehensive monitoring or control measures, putting critical business functions at risk.

Recommendations for Improvement

Respondents indicated a strong desire for improved observability and real-time transparency in their systems, with 62% emphasizing this as a critical need. Furthermore, 58% highlighted the necessity of demonstrating capabilities to integrate AI agents effectively into existing or core systems. Charles Crouchman, Chief Product Officer at Redwood, stated, _“The demo is the easy part. The real test begins when an AI agent interacts with production systems or customer transactions. At this point, governance cannot simply rely on policy documents but needs to be actionable within the process.”_

To tackle these challenges, further insights from Redwood’s research point to fundamental changes needed in organizational structures and strategies. This includes enhancing real-time monitoring capabilities and creating cohesive accountability frameworks.

A Call to Action

The report underscores the urgency for companies to reassess their approaches toward AI governance. With the alliance of AI systems to existing infrastructure representing the most significant barrier, there lies an opportunity for organizations to unlock potential efficiencies and safeguard business continuity. Developing actionable governance frameworks that merge strategy with operational execution can pave the way forward.

For those interested in diving deeper into Redwood's comprehensive findings, the full report, 'Enterprise benchmarks in AI and agentic orchestration 2026', is available for review. As businesses continue to untangle the complexities surrounding AI technologies, understanding these benchmarks will be crucial for informed decision-making and strategic planning.

To learn more about Redwood Software and its transformative capabilities, please visit Redwood's Official Website.