AI Agents are the Next Big Thing — Don’t Build Your Own, Advisors Say

AI agents gaining traction, but challenges abound: Agentic AI, designed to run specific functions without human intervention, is becoming increasingly popular among enterprises looking to automate workflows and leverage generative AI capabilities.

• Forrester has named AI agents as one of its top 10 emerging technologies for the year, highlighting their potential impact on business operations.
• However, the analyst firm warns that three-quarters of organizations attempting to build AI agents in-house will fail, citing the complexity of the process and the lack of specialized expertise in many companies.
• Forrester predicts that companies unable to build their own AI agents will turn to outside AI consulting firms or use agents embedded in software from their current vendors.

The complexity of AI agent development: Building AI agents requires advanced technical knowledge and resources that many organizations may not possess internally.

• AI agent architectures are described as convoluted, requiring multiple models, advanced retrieval augmented generation (RAG) stacks, and specialized expertise.
• The power of agentic AIs is still in its infancy, with analysts suggesting it may take another two years before they can meet “inflated automation hopes.”
• Organizations need a fully formed MLOps plan and the ability to integrate various open-source models into cohesive workflows.

Success stories and alternative approaches: Despite the challenges, some companies are finding success in developing their own AI agents or leveraging existing open-source models.

• Goldcast, a video marketing software developer, has experimented with a dozen open-source AI models to assist with various tasks, aiming to link them into autonomous agents.
• Slate Technologies began rolling out its own AI agents three years ago, even before the recent AI boom, demonstrating that in-house development is possible with the right expertise.
• These companies emphasize the importance of tailoring AI models to specific use cases and workflows rather than creating AI agents from scratch.

The importance of human oversight: Even with autonomous AI agents, human supervision remains crucial for successful implementation and ongoing improvement.

• Senthil Kumar, CTO of Slate Technologies, stresses that organizations can’t simply build an AI agent and forget about it; continuous monitoring and refinement are necessary.
• The development of AI agents is described as a collaborative process between the AI ecosystem and human counterparts, focusing on agent learning and knowledge dissemination.

The build vs. buy decision: Organizations face a crucial choice between developing AI agents in-house or partnering with external providers.

• Large companies may be tempted to develop highly customized agents but can face challenges related to fragmented internal data, resource underestimation, and lack of expertise.
• Chris Ackerson, head of AI at AlphaSense, suggests that buying solutions from trusted partners can help organizations avoid “builder’s remorse” and accelerate their path to success.
• Partnering with AI providers offers access to proven, ready-made agents that have been tested and refined, potentially saving time and resources.

The case for specialized expertise: Many experts advocate for leveraging external specialists or pre-built solutions to implement AI agents successfully.

• Adnan Masood, chief AI architect at UST, emphasizes the complexity of agentic AI architectures, including challenges like robust memory management and efficient search algorithms.
• Specialized expertise in machine learning, natural language processing, and data engineering is often required to build effective AI agents from the ground up.
• By turning to specialists or adopting pre-built solutions, organizations can benefit from the experience of those who have already navigated these challenges.

Looking ahead: The future of AI agents in business: As the technology matures, the adoption and implementation of AI agents are likely to evolve.

• While current limitations exist, the potential for AI agents to transform business operations remains significant.
• Organizations will need to carefully assess their internal capabilities and weigh the benefits of in-house development against the advantages of partnering with specialized providers.
• As the technology advances, we may see a shift towards more accessible and user-friendly AI agent development tools, potentially democratizing the creation of these powerful automation tools.