According to
There’s no doubt that generative AI is useful. That goes without saying. Generative AI can help with productivity in various different ways. However, they rely on “prompts” being continuously fed by humans to perform their tasks. On the other hand, AI agents don’t require as much oversight – they can carry out complicated tasks on their own. Instead of being a useful tool, AI agents serve more like collaborative teammates.
The Deloitte report highlights how AI agents can perform multi-step processes and respond to changes without needing constant human input. And the real potential lies in a multiagent approach.
A multiagent approach involves a system of multiple AI agents. Each agent within the system has a specialized skill and they all work together to carry out complicated workflows that would normally require human control at various points.
How does a multiagent approach work?
Let’s say your company writes weekly reports about weather events in a particular area. A multiagent approach might look something like this:
First, a “planning agent” starts by organizing the entire project and breaking it into smaller tasks. This agent then assigns specific tasks to other agents. One of the other agents collects weather data. This weather data is passed on to another agent specialized in writing summaries of the events, while another agent creates visuals. With the summaries and visuals ready, another agent pieces the writing with the visuals in a way and creates a report that focuses on an optimized reading experience.
Once the report is created, a “quality-check agent” then reviews it for accuracy.
It’s only when all the agents finish their tasks that a human analyst performs a quick check before publishing it.
Example of AI agents optimizing workflow
In this scenario, what would typically take the company days is done in less than an hour.
Deloitte’s report reveals how businesses are already using these multiagent systems in practical ways. They’re seeing real improvements in areas like financial advising, where agents can analyze vast amounts of financial data to give more precise recommendations. In retail, AI agents are used to adjust prices in real-time, personalizing offers based on market trends, inventory levels, and customer preferences.
Businesses should begin preparing AI agents
Companies should begin by looking for areas of the business where an AI agent could streamline processes. As noted in Deloitte’s report, it’s best to start with easier, lower-risk applications and then expand as teams gain experience with the technology.
For a multiagent AI system to be truly successful, companies should train their staff to collaborate effectively with AI. This could involve tweaking job functions to balance human and AI duties and promote a culture that sees AI…