AI agents are sparking a wave of excitement and debate. These advanced software entities do more than just process data — they grasp objectives, devise strategies, make autonomous decisions, and adapt through experience. And increasingly, they’re being explored as powerful tools in the fight against climate change; from managing energy use to optimising supply chains.

Naturally, their rise comes with both promise and risk. While the potential for positive impact is vast, so too are the concerns around autonomy, ethics and unintended consequences. That’s why staying informed (and asking the right questions) is essential if we’re to engage with this fast-moving technology in a way that feels grounded, balanced and safe.

What exactly are AI agents?

Imagine a digital assistant with significantly enhanced capabilities. Unlike traditional software that follows pre-programmed instructions, AI agents are designed to perceive their environment. This could involve interpreting customer data, understanding real-time sensor readings, or even processing natural language commands. Think of an agent that can “read” a complex financial report or “listen” to a customer support call and grasp the nuances.

AI agents are powered by foundation models like LLMs and are considered part of the next wave of generative AI. Think of it like this:

  • LLMs (Large Language Models), like GPT-4 or Claude, are a type of foundation model. They generate text by predicting what comes next in a sequence — super useful for writing, answering questions, summarising data, etc.
  • Generative AI is the broader category that includes LLMs, as well as models that generate images (like DALL·E), audio, or video. It’s the umbrella term.
  • AI agents are systems that use foundation models (including LLMs) as a core component, but go further than just generating content.

AI agents can reason, plan, remember, take actions, and interact with tools or environments to complete tasks with minimal human input.

Given a goal, an agent can break it down into logical steps, decide on the best course of action, and execute tasks independently — all while learning from experience. For example, an AI agent tasked with optimising a marketing campaign can analyse past performance data, identify target audiences, and plan a series of ad placements and email sequences.

As another example, it could help optimise energy use in a smart building by analysing weather forecasts, occupancy levels, and historical energy data, then making real-time adjustments to heating, lighting, and cooling systems.

The agent can interact with various systems; this could mean automatically adjusting inventory levels based on predicted demand, sending personalised customer emails, or even initiating a fraud investigation based on identified anomalies.

Its ability to learn and adapt

This part gets interesting. Through feedback loops and continuous learning mechanisms, AI agents refine their strategies over time. They analyse the outcomes of their actions and adjust their approach for better results in the future. An AI agent managing customer support might learn which responses are most effective in resolving specific issues.

They can remember past interactions and the current state of their environment, allowing for more coherent and effective long-term behaviour. An AI agent assisting with a multi-stage project can recall previous decisions and progress made.

Could AI agents help businesses meet net zero targets?

Many companies have set ambitious net zero goals — which is where AI agents could play a powerful role.

By constantly monitoring energy usage, supply chain emissions, and transport logistics, AI agents can identify inefficiencies and suggest alternatives in real time. Think of a logistics AI agent that reroutes delivery vehicles to minimise fuel consumption or selects lower-emission suppliers based on live data.

More advanced agents could even coordinate actions across departments — from facilities to procurement — helping align business operations with ESG targets automatically and at scale. And because they can adapt to changing conditions, these agents are well suited to navigating the complexities of decarbonisation in a volatile global market.

Can AI agents support a circular economy?

Yes — and this may be one of the most exciting frontiers. A key challenge in transitioning to a circular economy is keeping track of products, materials, and components as they move through multiple life cycles. AI agents could help manage this complexity.

Picture an agent embedded within a product lifecycle management system, tracking the condition and usage of parts in real time. When components near end-of-life, the agent could trigger collection or remanufacturing workflows, coordinate with recycling partners, or even redesign usage patterns to extend longevity.

In short, AI agents could help businesses shift from a linear ‘take-make-waste’ model to a regenerative one — not just by analysing data, but by acting on it autonomously and at speed.

How can business leaders get ready to embrace AI agents?

A report by McKinsey suggests there are a number of things businesses can do to prepare for the age of AI agents, including structuring their data and IT systems so that AI agents can interact smoothly with existing infrastructure. That means building in flexibility to adopt future technologies without causing disruption.

They also found that over 72% of surveyed companies are already deploying AI solutions, with growing interest in generative AI. As this momentum builds, it’s likely that more businesses will start factoring emerging technologies like AI agents into their strategic planning and future AI roadmaps.

Ultimately, the potential is huge, but so are the questions. How do we ensure AI agents act in line with human values, especially when making decisions that affect people or the planet? How transparent and accountable should these systems be? And what happens when autonomous decisions conflict with ethical or environmental goals?

These are the questions business leaders, policymakers, and technologists will need to grapple with in the months and years ahead. One thing is certain, however: AI agents are here — and if used wisely, they could become vital allies in building a fairer, more sustainable world.