.png)
Data-hungry AI agents are increasingly making purchasing decisions - and they want your Environmental Product Declarations. For businesses to remain competitive and achieve climate goals, we must be ready to serve these 'custobots', writes Anni Oviir.
It’s no longer impressive when companies say they have AI in their products.
I’d even say, as we enter 2025, it’s now become a bit of a red flag if a company is talking more about AI than their subject matter expertise or the actual value they offer.
Fortunately, as we emerge through AI’s hype cycle, we are increasingly seeing the real, tangible use cases for AI - borne of necessity rather than marketing.
Many of them are in our burgeoning cleantech sector.
Despite a fairly bumpy 2024 for cleantech startups, demand is continuing to grow for lower carbon products, propelled by fast evolving regulations across Europe and around the world.
AI is smoothing this green transition by powering cleantech that can help make sense of the big data around our carbon emissions and develop ever more intuitive ways to take the best actions with it. As with earlier SAAS products, some of the most promising applications of AI are first proving their worth as internal tools.
But I think we’re still missing the bigger picture here of how AI can contribute to the development of cleantech on a far more significant scale.
We shouldn’t just focus on the AI we can develop within cleantech. Every person and every business will have access to their own ever smarter AI. So we also need to think about how cleantech develops around everyone else’s AI. That’s how things get really exciting for cleantech in particular.
Procuring lower carbon materials is incredibly complex when, for example, designing and constructing a building. In fact, it’s just one additional complex factor among many others. You’ve got to weigh up the embodied carbon, alongside variations in material durability, its fire resistance, noise performance, thermal performance, and many other factors, not to mention the financial cost too.
But this is exactly the kind of complexity that AI is rapidly learning to solve.
Rise of the ‘custobots’
It was Gartner, the tech research consultancy, that pioneered our understanding of hype cycles in technology, as we are now seeing play out with AI. Pay close attention to a more recent Gartner prediction though, specifically related to AI. They predict that, just as e-commerce emerged as a megatrend of the internet, machine buying will emerge as a megatrend of AI.
Machine buying is essentially when AI agents help us do our shopping - or our company’s procurement. Humans set the requirements and parameters, and keep control of the key decisions, but it’s AI agents that take care of the hassle when it comes to finding and negotiating the best deals.
This isn’t hypothetical. Many of the world’s largest companies, such as Walmart, already use AI agents to autonomously negotiate and conclude billions of dollars in deals with their suppliers. Walmart is scaling the technology across its business and other Fortune 500 companies, especially in manufacturing, are following their lead.
Companies using this technology have discovered that machine buying enables them to make better purchasing decisions by weighing up far more complex factors than ever before possible.
It doesn’t just get them better deals done faster. It also gives them the capability to use their purchasing power in new ways to achieve bigger goals.
Many of them have discovered that AI-powered buying enables them to more easily achieve corporate social responsibility goals, such as buying more from local family businesses.
And it won’t just be the Fortune 500 driving this. The most widely-used AI tools, such as ChatGPT and Google Gemeni, are in a race to roll out agentic AI shopping features too.
As we all start to make much more efficient and timely purchases with the help of AI, machine buying on a wider scale could unleash a wave of economic growth on an unprecedented scale.
That’s according to Don Scheibernreif and Mark Raskino, senior researchers at Gartner, who have since published When Machines Become Customers to help businesses prepare to serve AI agents - or ‘custobots’, as they call them.
Custobots will be ruthlessly efficient, they warn, so they won’t be swayed by existing marketing and sales tactics that focus on human persuasion, often based on emotions or biases. Instead, they’ll hone in on the precise data they need to make the best possible purchasing decisions. It means they won’t care about a clever slogan or fancy packaging. In fact, if lowering carbon emissions is part of their purchasing goals then they’ll be seeking the data of how much embodied carbon is in that packaging, as well as the rest of the product.
As Scheibernreif and Raskino warn, all companies will need to radically rethink how they market and sell in order to serve AI agents representing customers.
Afterall, you can’t take an AI out for dinner but you can feed it data. To remain competitive, companies will need to ensure they have the kind of data that AI agents like to consume in order to make their purchasing decisions.
Greenwashing won’t work. The AI agents will skip past the fluffy claims and go straight to the hard data that’s credible, transparent, and third party verified in line with common standards.
Fortunately, we already have Environmental Product Declarations (EPDs), which fit that purpose and will become increasingly important as we enter this age of AI-led buying.
Why AI agents love EPDs
An EPD is a standardised and verified way of displaying environmental information about a product based on its life cycle assessment. It helps anyone involved in purchasing decisions to make comparisons between competing products. They are used internationally, subject to complex variations in rules and regulations, and their value for manufacturers is growing.
EPDs are especially important in construction, which accounts for around 40% of global carbon emissions, much of which is embodied in the production of materials.
Fortunately, the construction sector is also at a more advanced stage of accounting for its carbon impact as regulations around carbon limits, determined by the availability of EPDs, come into force across Europe and around the world. In some markets, it’s not even possible to participate in certain tenders with an EPD. The contractors need to ensure that the entire building fits within embodied carbon limits based on the sum total of its materials.
The EU has just recently passed the Construction Products Regulation that will further accelerate this across the continent.
Manufacturers that reduce their carbon footprints - and get that verified - can increase their competitiveness. Just as importantly, a good quality EPD with high standard verification helps identify how best to lower carbon emissions even further in future.
It is particularly auspicious to have data-driven machine buying emerging as a significant use case for AI at the same time that we face requirements for verified data on carbon emissions.
It means the construction sector is likely to be the trailblazer in demonstrating to other sectors - such as in agriculture, textiles, and automotive - just how well machine buying can improve efficiency while also contributing to climate goals.
That’s the big opportunity for the development of cleantech.
So although machine buying is not being developed explicitly for the purpose of cleantech, it will be the AI that will have the most profound impact on cleantech.
But we can’t sit back and expect any of this to develop inevitably. We must start preparing now.
What this means for businesses
For those of us working in cleantech, we should be humble when we talk about developing our own AI and not lose sight of the fast evolving landscape around us that we aim to serve.
Preparing to work with other people’s AI is just as, if not more, important as developing our own.
There’s certainly still huge value to be gained as we continue to develop our own AI tools in cleantech for ourselves and our clients to work more efficiently. But there’s also exponentially greater value from broader-purpose AI tools in the hands of others and we must think more deeply about how we adapt to this.
Most critically, we must accelerate the digitisation of the EPD process. That’s not just about making it easier for those of us specialised in EPDs to work and publish online. It’s also ensuring all EPDs are machine readable so that buyers - or their AI agents acting for them - can easily access the embodied carbon of different products as part of their procurement process.
All companies, regardless of sector, whether B2B or B2C, must now think about adapting to the data hungry AI that will increasingly be used by customers.
We’ve been through this kind of change before with the internet revolution, which meant businesses had to radically adapt to things like search engine optimisation and social media engagement. We must now adapt again to the AI revolution.
With just as much at stake, businesses will increasingly need to ensure they have transparent and credible data available for their customers, both human and AI.
Thanks for reading. If you’d like to hear more about our own cleantech journey, follow EPD Verifier on LinkedIn or drop us your email at EPDVerifier.com.
