Agentic AI, Sovereign AI, and What's Coming — A Leader's Horizon Scan

The Meridian Story

In a strategic planning session, Sarah (CEO) asked the team a forward-looking question: "What should we be watching that isn't on our current roadmap? What might be table stakes in two years that we'd want to be prepared for?"

This kind of horizon scanning isn't speculation — it's strategic preparation. Capabilities that seem distant today often become operational expectations faster than expected. Leaders who track emerging trends can position their organizations to adapt rather than react.

Trend 1: Agentic AI

What it is: AI systems that can plan and execute multi-step tasks with a degree of autonomy. Instead of answering a question, an agent might research a topic, draft a document, schedule meetings, send communications, and follow up — all from a single high-level instruction.

Where it stands today: Early enterprise adoption. Agentic capabilities exist in commercial tools, but production-grade enterprise deployment is still maturing. PwC's 2026 predictions identify agentic workflows as a significant emerging opportunity, with the caveat that organizations need clear governance and human oversight (PwC 2026).

Strategic implications:

• Agents can handle workflows that span multiple systems and require sequential decisions — areas previously difficult to automate
• The data demands for agentic systems are even higher than for supervised AI: agents need real-time, accurate, well-governed data to make autonomous decisions
• Governance becomes more critical, not less. Autonomous action requires robust oversight frameworks
• Roles evolve: humans shift from doing the work to designing and supervising agent workflows

What to watch: Track how peers in your industry are piloting agentic capabilities. Identify 1–2 workflows where agentic AI might be valuable — typically multi-step processes that span systems and require coordination.

Trend 2: Multimodal AI

What it is: AI systems that work across multiple types of input and output — text, images, audio, video, structured data — in unified ways. A multimodal system might analyze a video, generate a written summary, create accompanying charts, and answer follow-up questions about all of it.

Where it stands today: Rapidly maturing. Major AI providers offer multimodal capabilities. Enterprise applications are emerging, particularly in fields involving rich media (manufacturing, healthcare imaging, media, retail).

Strategic implications:

• Many enterprise data sources have been underutilized because they're in non-text formats: meeting recordings, security camera footage, manufacturing imagery, training videos
• Multimodal capabilities can unlock value from these previously hard-to-analyze sources
• Data infrastructure expands to include unstructured media, not just text and structured data

What to watch: Identify high-volume non-text data your organization generates that might benefit from multimodal analysis (call center recordings, video communications, visual records).

Trend 3: Sovereign AI

What it is: AI capabilities that organizations or governments control end-to-end — running on their infrastructure, using models they control, with data that doesn't leave their environment. The driver: data sovereignty, regulatory compliance, and reduced vendor dependency.

Where it stands today: Growing interest, particularly in regulated industries (financial services, healthcare, government) and in regions with strong data sovereignty regulations. Open-source models have made sovereign AI more practical than it was 2 years ago.

Strategic implications:

• For regulated industries or sensitive use cases, sovereign approaches reduce compliance and dependency risks
• Open-source models combined with private cloud or on-premise infrastructure offer alternatives to public AI APIs
• Trade-off: sovereign AI typically requires more internal capability to deploy and maintain
• Hybrid approaches (sovereign for sensitive use cases, commercial APIs for others) are common

What to watch: Assess which of your AI use cases would benefit from sovereign deployment based on data sensitivity, regulatory requirements, and strategic importance.

Trend 4: Physical AI

What it is: AI integrated with physical systems — robotics, autonomous vehicles, smart manufacturing, IoT-enabled operations. AI capabilities extend beyond information processing into physical action.

Where it stands today: Maturing rapidly in specific domains. Manufacturing automation, warehouse robotics, autonomous vehicles in controlled environments are operational. General-purpose physical AI is still developing.

Strategic implications:

• For organizations with physical operations (manufacturing, logistics, field services), physical AI represents significant operational opportunities
• Integration of digital and physical AI requires investment in sensors, connectivity, and data infrastructure
• Safety and reliability requirements are different from purely digital AI

What to watch: For organizations with physical operations, identify where AI-enabled physical systems could improve safety, quality, throughput, or cost.

How These Trends Amplify Data Readiness

A consistent pattern across all four trends: data readiness becomes more important, not less.

• Agentic AI needs real-time, accurate data to make autonomous decisions
• Multimodal AI expands data infrastructure to handle rich media
• Sovereign AI requires complete data control end-to-end
• Physical AI generates and consumes massive volumes of operational data

The data foundations built today (Lessons 8–9) become more valuable as these trends mature. Organizations with strong data readiness will adopt these capabilities faster and more reliably than those without.

What's Hype, What's Real

It's useful to apply healthy skepticism to emerging trend coverage:

Often Overhyped	Often Underappreciated
"AI will replace [profession] within X years"	Compounding value of foundational work
Specific timeline predictions	Importance of organizational adoption capability
AI capabilities demonstrated in narrow demos	Engineering effort required for production reliability
Universal applicability claims	Domain-specific advantages

A practical approach: track multiple credible sources, focus on what enterprises are actually deploying (not what's being demonstrated), and weight peer industry adoption more heavily than vendor projections.

What This Means for Your Organization

• Horizon scanning is part of strategic leadership, not a separate activity. Allocate time quarterly to track emerging trends.
• Avoid both extremes: dismissing emerging capabilities as hype, or rushing to adopt before they're production-ready for your context.
• Foundational investments — data readiness, governance, talent literacy — make adoption of any emerging capability faster and more reliable.
• Identify 1–2 trends most relevant to your industry and operations, and track them more deeply.

Key Takeaways

• Four trends matter most for enterprise strategy: agentic AI, multimodal AI, sovereign AI, and physical AI.
• Each trend has higher data readiness requirements than current AI applications. Foundational data work compounds in value as these trends mature.
• Apply healthy skepticism to trend coverage: focus on what enterprises are actually deploying, not what's being demonstrated.
• Identify 1–2 trends most relevant to your context for deeper tracking; for the rest, monitor at a high level.
• Foundational investments — data, governance, literacy — position the organization to adopt any of these trends faster than starting from scratch.

Next Lesson

The course concludes with a practical action plan. In Lesson 26, we'll build your first 90 days as an AI-ready leader — a concrete, week-by-week plan to apply what you've learned, regardless of where your organization is on its AI journey.