• December 10, 2024
Report Says AI Agents will Outpace LLMs for Automating Complex Workflows

A new report from consulting giant Deloitte suggests that AI agents—a new technology that many predicts will dominate business process automation moving forward—will be able to handle more complex workflows than generative AI.

According to Prompting For Action: How AI Agents are Reshaping the Future of Work, while Gen AI has already led to significant efficiency gains for businesses by producing novel outputs in response to plain language prompts, AI agents will supercharge those gains due to the ability to understand context, plan workflows, connect to external tools and data, and execute actions that progress toward a defined goal.

The report notes that the LLMs generative AI is based on (and small language models many organizations are turning to because they are quicker and less costly) have difficulty understanding multi-step prompts and reasoning outside a specific task.

“AI agents excel in addressing these limitations while also leveraging capabilities of domain- and task-specific digital tools to complete more complicated tasks effectively,” the report’s authors wrote. “For example, AI agents equipped with long-term memory can remember customer and constituent interactions—including emails, chat sessions and phone calls—across digital channels, continuously learning and adjusting personalized recommendations. This contrasts with typical LLMs and SLMs, which are often limited to session-specific information. Moreover, AI agents can automate end-to-end processes, particularly those requiring sophisticated reasoning, planning and execution.”

The report examines, in detail, the characteristics that make agents superior to LLMs and SLM, use cases toward which agents can be applied and what might happen in the future. Deloitte acknowledges that AI agents are a very early stage technology, but recommend that businesses begin to assemble a technology infrastructure that will support the adoption and implementation of AI agents, ensure their data is properly organized and accessible to agents and that adequate governance is in place to monitor and manage performance and potential bias.