What is an agent?

Nowadays, everyone is calling everything an agent, a co-pilot, or an assistant.

So how do we talk about these things with precision in order to make decisions about what to invest in?

The best way to do this is to think through what they actually do on behalf of your employees or customers in relation to decisions.

Below is a framework for this that should help.

There are 26 categories of AI, of which Generative AI is only 1. A system’s capability to answer questions, guide through a process, perform research, or act on one’s behalf is made possible by orchestrating different categories of AI, where relevant, together with software.

Sources and Analysts (Naked GenAI)

ChatGPT is an example of a source and a rudimentary analyst. I say rudimentary because its analytical capabilities are still limited. ChatGPT is not an agent you would want to give your credit card information to and assign a marketing campaign. Neither is it a trainer you would trust to teach a new hire how to do their job at your organization.

Creating your own GPT in ChatGPT is like creating a customized source and analyst. But from a technical standpoint, you are essentially adding whatever expertise you upload as a topping on a generalized GPT pizza.

The majority of Generative AI solutions out there today fall into the source and analyst category, such as Perplexity, Microsoft Co-Pilot, Claude, etc.

Trainers (GenAI + Indices + Expert Systems)

I can’t tell you about the best example of a trainer I’ve seen other than to say that it has video content of experts performing advanced processes that new hires can interact with Generative AI to find, but Generative AI is being used as the “intent finder” and not to supply any answers.

Agents

Agents have multiple definitions. According to an overwhelming number of online sources, an LLM agent is: “an artificial intelligence systems that utilizes a large language model (LLM) as its core computational engine.”

👎🏼

Here’s how I define it:

An AI agent is a system that can perform tasks and make decisions on your behalf.

Technologists keep trying to explain how they work, and they really like this thing called “Multi-Agents."

There are two ways people are talking about multi-agents.

Here’s the good way:

A network of agents who are each independently capable of performing tasks and making decisions on behalf of an organization.

👆🏼 this is an extremely promising new field of research and work, with the potential of enabling cross-organizational negotiations and trade-offs at lightning speed.

Here’s the bad way:

LLM multi-agents is when you have a scripted series of prompts that interact with each other.

For example, if your goal is to create a system that can generate a really great sales pitch, you might have the “Idea Agent” followed by the “Narrative Agent” followed by the “Pricing Agent", and so on.

This is automation, not autonomous. You’ve predetermined how the process will flow and prompted a series of scripts. It’s helpful for conceptualizing an LLM flow, but calling it agents is inaccurate and creates confusion. It’s more like automation with LLMs.

So who’s building real agents?

Rabbit R1 and Humane AI have both attempted to create consumer agents, though there are limitations to both products thoroughly covered elsewhere. My bet for where we’ll see the best real-world consumer agentive capability is Apple.

There are two companies building enterprise agents that I’ve been really excited about (neither of which have sponsored this newsletter—this is my objective opinion):

Aera Technology has a vision for the autonomous enterprise that I find quite compelling. They don’t describe their work as agentive, but they have created the capability for operations leaders, supply chain planners, etc. to teach their system how they prefer to make decisions and assign that the system can autonomously make any decision it feels x% confident about, handling, in many cases, millions of dollars’ worth of decisions per day.

Composabl is the only agent-building company I’ve seen that extends to the factory floor and is the most flexible given that it is a workbench rather than a point solution. Fortune 500 companies are using their system to build agents to autonomously operate heavy machinery at lightning speed. Full disclosure that I am a Strategic Advisor for them, but I’ve only accepted this position because of how bullish I am about this capability and their founder.

I am also aware of a few startups that are building enterprise agents, some of which have not introduced a product yet, but I will certainly keep you updated when they do.

*If you know of any products/companies I’m missing in any of these categories, especially in the agent category, please don’t hesitate to comment or reach out to share it with me.

Thanks for reading,

Brian

---

Whenever you're ready, here are 3 ways I can help you:

1. My LinkedIn Learning Course launched on October 3rd: Organizational Leadership in the Era of AI. This 48-minute course is packed with frameworks and insights to help you lead in the era of AI using the new system of leadership I introduced in Autonomous Transformation and is free to anyone with a LinkedIn Learning subscription.

2. If you want to go deeper as an individual, you can sign up for my live course Future Solving Fundamentals. My flagship live course on how to position yourself in the future in the era of AI. I share over a decade of AI strategy expertise, proven methods, and actionable strategies. This course sets the stage for a new era of value creation with artificial intelligence. Join leaders from Microsoft, Accenture, Amazon, Disney, Mastercard, IKEA, Oracle, Intel, and more.

3. If you want to go deeper as a team or organization, I’ve partnered with LinkedIn Learning, Emeritus, and Duarte to create tailored Future Solving™️ programs that leaders at Fortune 500 companies are implementing to help reembrace science in their decision-making and meet the complexities of the 21st century with a system of leadership, strategy, and decision-making that is built-for-purpose for era of artificial intelligence. You can reach out to connect and learn more here.