Boosting Business Value with AI Agents Key Strategies

Introduction

Agentic artificial intelligence (AI) is the latest wave of solutions transforming how organizations operate and compete. AI agents promise to enhance resource efficiency, automate complex tasks, and drive business innovation beyond traditional automation tools. However, deploying AI agents at scale comes with significant challenges, including unrealistic expectations and technical complexities.

Current Landscape

According to a January 2025 Gartner poll of 3,412 webinar attendees:

• 53% of organizations are in exploration mode with agentic AI.
• 25% are piloting AI agents.
• Only 6% have deployed them in production.
• 40% plan to pursue initiatives within the next six months.

Despite the hype, many organizations struggle with the gap between expectations and operational reality, particularly when building AI agents on top of large language models (LLMs). These solutions often lack the contextual understanding required for enterprise decision-making.

Key Strategies for Success

1. Mix and Match AI Agent Capabilities

AI agents excel at understanding user intent, retrieving and processing information, and leveraging tools to complete tasks. Organizations should adopt a flexible approach by:

• Configuring or building agents based on available data.
• Integrating tools and systems they need to interact with.
• Leveraging LLM capabilities tailored to their business context.

2. Understand Limitations

AI agents have critical limitations, including:

• Lack of World Models: Unlike humans, AI agents cannot build internal representations of their environment to predict outcomes. Current memory components rely on chat history and system logs, which fall short of capturing dynamic world states.
• Correlation vs. Causation: LLMs learn from data distribution, identifying correlations rather than causations. For tasks requiring accuracy (e.g., route planning), graph-based algorithms outperform LLMs.
• Composite Systems: AI agents are not standalone models but composite systems combining multiple techniques for perception, reasoning, and action.

3. Focus on Core Enterprise Components

Given the rapid evolution of AI agent technology, organizations should:

• Adopt an agile approach to minimize latency and build trust.
• Prioritize "plug and play" components to avoid vendor lock-in.
• Leverage open-source AI agent stacks or vendor solutions that promote interoperability.

Conclusion

While AI agents hold immense potential, organizations must navigate their limitations and adopt strategic, agile approaches to unlock business value. By mixing and matching capabilities, understanding constraints, and focusing on core components, businesses can position themselves for success in the evolving AI landscape.

Related Reading

• Lloyds and Nationwide to use UK finance sector LLM
• The role of small language models in enterprise AI
• Are LLM firewalls the future of AI security?

Ben Yan is a senior director analyst at Gartner focused on data and analytics.