In a previous article, I explored why the AI revolution demands a new “Control Plane” to manage the chaos of ad hoc automation. Today, that revolution is accelerating. We are rapidly moving beyond single-task models to a future dominated by agentic AI—complex systems where multiple autonomous agents collaborate to achieve sophisticated goals.

The shift is happening faster than many anticipate. A September 2025 Dimensional Research report sponsored by my organization found that 98% of companies plan to deploy agentic AI within the next 12 months. This momentum is reflected in broader markets. Gartner forecasts worldwide AI spending to total $2.52 trillion in 2026, a 44% increase year-over-year.

However, without a structured framework, the very autonomy that makes agentic AI so powerful creates a new tier of operational risk. To unlock agentic AI’s potential, organizations must evolve their Control Plane strategies to address two critical realities: the volatility of data and the necessity of holistic orchestration.

The Unseen Anchor: Why Bad Data Kills Agents

Brittle data pipelines are a nuisance for standard analytics; for agentic AI, they are fatal.

Goal-oriented agents are voracious for high-quality, timely data. When an autonomous agent encounters outdated or incorrect information, it often executes transactions or modifies infrastructure based on false premises.

The core challenges—ensuring data quality and handling massive volumes—are compounded by data locked in disparate systems like ERPs and mainframes. The result is that many data experts spend a disproportionate amount of their time managing a complex web of data pipeline tools.

In fact, according to a report by CData, 71% of AI teams spend more than a quarter of their implementation time on data integration, including modeling data, implementing ETL pipelines and configuring connectors.

Data integration represents a significant hurdle for most organizations in their agentic AI journeys. Last year, Capgemini surveyed 1,500 executives at organizations with more than $1 billion in annual revenue, across 14 countries. Only 9% of organizations reported that they are fully prepared in terms of data integration and interoperability.

A Holistic Framework For Orchestration

A centralized orchestration layer and strategy is crucial for addressing these challenges. Without it, organizations face several operational risks:

• Workflow Breakdown: Agents competing for API rate limits or processing power can crash mission-critical workflows.

• Loss Of Visibility: Without centralized monitoring, tracking why thousands of agents made specific decisions becomes a forensic nightmare.

• Security Breaches: When agents aren’t orchestrated and operate outside established rules, they create vulnerabilities and make GDPR or HIPAA adherence a monumental challenge.

Just as autonomous vehicles require traffic signals, agentic AI requires an orchestration framework. IT leaders recognize this: Ninety-three percent of companies plan to implement an orchestration solution to manage their agents. Platforms can help with integrating agents with legacy systems via API management, comprehensive observability, robust governance and advanced analytics.

While a robust orchestration platform is a critical first step, this layer achieves its full potential only when integrated into a broader strategy across three pillars: People, process and technology.

1. People: The Emergence Of The “Agent Strategist”

Successful organizations will need a new class of professional: The Agent Strategist.

As agents handle autonomous tasks, human roles are shifting toward high-level orchestration. Agent Strategists act as “air traffic controllers,” monitoring agent health, tuning performance and guiding autonomous logic.

This allows experts to focus on intent rather than manual entry. Upskilling now focuses on “strategic oversight” and “algorithmic risk management” to keep humans effectively in the loop.

2. Process: Optimizing For Autonomy

Aligning processes with agentic capabilities ensures automation drives meaningful efficiency.

Companies must establish “rules of the road”—i.e., security boundaries and compliance limits—before translating them into platform configurations. They must also define trigger events for high-stakes actions where the process must pause for human validation.

Along the way, auditing workflows to ensure agentic interactions are complementary is crucial to prevent resource conflicts and system-wide instability.

3. Technology: Scalable Infrastructure And Identity

Finally, the surrounding infrastructure must treat agents as first-class citizens.

Assigning distinct identities to each agent, for instance, allows for least-privilege access and absolute accountability via digital signatures. Companies must also prioritize “reasoning traces”—immutable records of why an agent made a decision—to satisfy regulators and simplify diagnosis.

When building out their new stack, one consideration is that infrastructure must be elastic to handle “bursty” sequences of API calls without impacting legacy system performance.

Conclusion

The Control Plane is no longer just a defensive measure; it is a strategic offensive capability.

By planning for orchestration in parallel with AI development—and focusing on foundational data—organizations gain the control needed for success. This approach can transform data management from a manual burden into a centralized, automated process, ensuring that while our agents act autonomously, they never act out of control.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?

Source link