American semiconductor maker AMD has proposed creating a new category of computer to cater to the rise of artificial intelligence.

In a recent blog post, the tech firm explores how the rapid advancement of AI technology could result in so-called agent computers replacing the personal computer that humans have come to rely on over the last few decades.

Whereas personal computers require humans to sit in front of a screen and control it using a mouse and keyboard to complete tasks like sending emails and browsing the internet, AMD says an agent computer would require little input from users.

Agentic artificial intelligence systems, which are designed to complete tasks autonomously in accordance with human user aims and objectives, would instead do all the heavy lifting. AMD uses the example of a human user creating a task in Slack for the agent computer to complete, or sending a message to their computer asking for a status update and the computer replying with a detailed report.

The difference between personal and agent computers is perfectly summarised by AMD in the blog post. “A personal computer runs your apps,” writes the firm. “An Agent Computer runs your agents so they can run the apps for you.”

Elsewhere in the blog, AMD provides a glimpse into different real-life scenarios of agent computers and their impact on users. One of them is a person's agent computer creating a list of daily priorities, replying to important messages and creating a meeting briefing document before they have woken up.

In a second scenario, AMD suggests that an agent computer could pull together several pieces of research to help someone complete a project they are currently in the middle of. They would just need to tell the computer what research is needed and how it should be viualised, like creating a written summary of competitive landscape data. They can then go about their day and return to the computer to find the “finished deliverable”.

Such computers would represent a big change in how AI workloads are currently handled. As opposed to being stuck in a faraway hyperscale data center, some AI workloads would instead be stored locally on the user’s own agent computer. AMD suggests this could improve data privacy, ensure AI agents cater to users’ individual needs and make this technology more accessible for everyday consumers.

But given that agentic computers would operate around the clock and AI tools known to be power-intensive, AMD knowledges this new class of computer would only be possible with “powerful hardware, high-bandwidth unified memory, efficient parallel compute, and the architecture to run sophisticated AI models”.

AMD concludes the blog by pointing out that it’s already working to make agent computers a reality. The company believes that its Ryzen AI Max+ processors, such as the Ryzen AI Max+ 395, are capable of acting as agent computers.

It says: “They have the horsepower to run sophisticated local models, the efficiency to stay persistently available, and the architecture to support the parallel, multi-agent workloads that define what comes next.”

---