Multiplayer Agents: The Next Wave of Collaborative Agentic AI

Highlights

• Multiplayer agents are the new frontier of the AI revolution. 
• This allows for a shared context between multiple agents and humans in a single conversation.
• Deep collaboration between humans and AI could be the key to solving the AI adoption problem that leads to a lack of positive ROI in AI projects. 

First, there were AI chatbots. Then, we got AI assistants. Next, we saw the rise of the agentic enterprise. The latest trend is the emergence of multiplayer AI agents. Almost every major AI vendor has announced their own version of a multiplayer agentic system, and it’s a trend that I don’t think we’ll see slowing down anytime soon.

The question remains: What exactly is a multiplayer agent? Why is it important, and what is the benefit that it offers above regular agentic AI? Another, perhaps more important question, is: Are businesses going to see an ROI on this technology, or is it just another AI money pit? All this, and more, will be addressed in this article.

Defining “Multiplayer Agents”

To define multiplayer agents, we first need to identify what a single-player one is. These are the AI agents that are accessed by one user and one or more agents, through one chat (or other interface), and create a single thread of context (a session) that dies out at the end of the conversation. 

These sessions may be saved for review later, or to be used as memory in future conversations or tasks, but these sessions are not open to other users to contribute to.

This is where multiplayer agents come in. A shared context is created, and multiple users and agents can contribute to it simultaneously. A single context, but shared with multiple humans and agents. The context persists throughout, the agent reasoning is visible to everyone in the conversation, and multiple agents can take their specific actions in parallel. 

It seems like many vendors are moving away from the days of a single chat interface and adding “group chat” functionality to their products. They know that human beings work in teams and that silos can stifle productivity, and they’re looking to solve that problem when it comes to AI agents.

Why Single-Player AI Stalls

While I’m the first to get excited about new AI innovations, I’m also not going to neglect to point out that businesses are still struggling to see a return on investment from their AI projects. A recent report from the Atlanta Federal Reserve found that more than 90% of executives have not yet found value from their AI projects in terms of a productivity boost. This is not a small fraction of the market!

So, the question is no longer if AI is returning value to the business that invests in it, but rather why isn’t there a more universal return on investment? We know that AI was blamed for a large number of layoffs over the last few years, so human beings have lost their jobs as a result. 

The assumed intent behind these layoffs was that AI would pick up the slack and maintain or increase productivity, yet these numbers don’t seem to agree with that.

According to the report, the biggest reason that AI isn’t returning value isn’t due to the technology itself, but rather the lack of adoption of said technology. There’s no denying that many people are hesitant to adopt AI, and this seems to be driving the negative impact on the overall return on investment.

You may have heard the old marketing adage that you can have the greatest product in the world, but it’s no good if no one knows about it. The same can be said for AI. You can have the best AI implementation in the world, but if no one is using it, it’s not going to be very beneficial. It looks as though it’s not the technology itself that’s causing the issue, but the resistance to change, and specifically the hesitancy to adopt artificial intelligence tools.

Could Multiplayer Fix the Bottleneck?

One of the biggest complaints about AI tooling is how impersonal it feels, and how it may be having a negative impact on how humans interact with each other. I genuinely wonder if multiplayer agents could help with this problem. Instead of humans being told to work in silos and potentially seeing the AI as a replacement for another human being, multiplayer agents could provide a more interactive environment. 

The other thing to note is that this way, agents aren’t operating in the shadows. They are reasoning and taking action out loud, in the presence of their human teams. These teams can then see a history of the actions taken by the agent, the sources of data, and any limitations that the agent is facing. It also gives the humans an opportunity to step in where required and pull the correct individual or agent into the conversation.

When working with other humans, there’s nothing worse than having to explain a problem to someone who wasn’t in the room when it was first discussed. The same goes with agents. Multiplayer agents solve this problem as well by ensuring that the context is shared between all of the agents and humans in the conversation. No more repeating the same information over and over again; it’s just there in the thread.

Humans and agents alike, working together in a collaborative way. I am hopeful that this could begin to solve some of the adoption issues that are currently plaguing AI investments.