Why your business needs an AI agent strategy

AI agents poised to drive ROI from artificial intelligence: The emergence of AI agents is set to become a major trend in late 2024, offering businesses a practical way to leverage AI technology and measure its impact.

• AI agents are composite, autonomous applications that enable collaboration between humans and AI to complete tasks, going beyond the capabilities of existing AI-assisted copilots or chatbots.
• These agents can take actions based on LLM responses, with or without human involvement, and comprise multiple reasoning or action capabilities.
• Unlike chatbots, AI agents can be given specific workflows or tools to complete activities, making their performance measurable against business KPIs.

Key advantages of AI agents: AI agents offer several benefits over traditional AI applications, making them particularly valuable for businesses seeking tangible returns on their AI investments.

• Purposeful design: Agents are created for specific workflows, unlike the freeform nature of copilots and chatbots, allowing for more precise measurement of their impact.
• Accessibility: Non-developers can create and manage agents using no-code or low-code tools, democratizing AI implementation within organizations.
• Contextual responses: By leveraging predefined LLM prompts and well-defined services and data, agents can provide more accurate and contextual responses than public AI chatbots.
• Integration capabilities: Agents can seamlessly integrate with existing databases and systems, enhancing their ability to make informed decisions and recommendations.
• Persistence: AI agents can remember preferences over time, improving their performance and user experience with continued use.
• Ambiguity management: The natural language interface allows agents to handle uncertain situations more flexibly than traditional deterministic applications.

Technical components of AI agents: Understanding the structure of AI agents is crucial for businesses looking to implement this technology effectively.

• Large Language Model (LLM): Serves multiple roles, including guiding processes and making reasoned decisions based on inputs.
• Data: Includes application data for programmatic elements and inference data from the LLM.
• Governance: Comprises business rules and parameters that constrain the agent’s actions.
• Integrations: Elements of connectivity and data provenance among different applications and services.
• Workflows/Rules: Guide the agent and LLM through processes.
• Human Interface: Facilitates collaboration between the agent and humans.
• Persistence: Stores data of interest to the agent for knowledge injection and inference shaping.

Projected timeline for AI agent adoption: The rollout of AI agents is expected to occur in several phases throughout 2024 and 2025.

• Current phase: Application platforms like Salesforce and ServiceNow are leading the way in agent deployment.
• Early 2025: Infrastructure platforms such as AWS and Oracle Cloud Infrastructure will introduce more technology-agnostic offerings.
• Early to mid-2025: AI startups are likely to pivot their strategies towards agent development, potentially creating new business models and marketplaces.
• Mid to late 2025: First-of-its-kind production deployments are expected to emerge, showcasing real-world applications of AI agents.

Challenges and considerations: As with any emerging technology, AI agents face several hurdles that businesses must address for successful implementation.

• Governance: Ensuring proper security and policy application for the data used by agents is crucial for managing risk.
• Testing: Developing effective testing methodologies for non-deterministic AI systems presents a unique challenge.
• Observability: Determining appropriate metrics and expectations for agent performance will be essential for effective management.
• Performance and efficiency: Thorough testing and piloting are necessary to identify and address potential bottlenecks or unforeseen issues.

Recommendations for enterprises: To capitalize on the potential of AI agents, businesses should consider the following steps over the next 12 months.

1. Start with pilot projects using existing application platforms to understand organizational and technical gaps.
2. Designate a “general manager of agents” at the C-level to oversee implementation and manage associated risks.
3. Prioritize extensive testing to establish best practices and measure success in the absence of established industry standards.

The impact of AI agents on business operations: As AI agents continue to evolve, they are likely to play a significant role in helping organizations realize the practical benefits of AI technologies.

• The ability to leverage organization-specific data and services will enable AI agents to operate more precisely and add more value than generic AI solutions.
• The integration of AI agents with established application platforms may provide the entry point many executives have been seeking to justify AI investments.
• The next 12 months are expected to be a critical period for businesses to explore and understand the potential of AI agents in driving measurable ROI from their AI initiatives.