Earlier this month, Sam Altman described the appointment of Peter Steinberger at OpenAI as one that would “drive the next generation of personal agents”. As the founder of $500 billion agentic AI start-up, OpenClaw, Peter’s appointment appeared to signify a meaningful push towards a more widespread adoption of agents, building on the generative AI of ChatGPT. Created in just one hour, OpenClaw had already made 1.5 million AI agents by the start of February which were being used by individuals to undertake time-consuming tasks like managing emails and calendars through apps such as WhatsApp, Slack and iMessage. It was a viral sensation and highlighted a level of unprecedented access and productivity.
Although agentic AI is only just beginning to breakthrough to the general populace’s consciousness, talk of this technology has spread quickly across retail and consumer businesses. At the end of last year, two behemoths, Walmart and Google, announced a partnership which was centred around utilising agents on Google’s Gemini platform. Gemini would now be able to discover Walmart and Sam’s Club items when performing research and, if users decided they wanted to buy something, they could checkout and decide on shipping options without leaving the chat.
These developments although small give a flavour or what consumers can expect to see in the near future. I spoke to a senior retail technology leader, who described to me the current moment in practical terms. “In terms of agents, there’s a lot of hype ahead of reality, there’s no doubt about that, however things are moving at real pace, and the next 12 month will seismic shifts in adoption and use cases,” he said. “You will start seeing agents actually working across systems and doing things reliably and autonomously, that’s when it starts to get real.” The examples may still be limited, but they are enough to convince many organisations that this is something to seriously take note of.
Customer interaction is likely to provide the clearest early test. Search and shopping still account for most direct contact between a consumer business and its customers, which makes them a natural place to experiment with delegated decisions. This boundary is likely to move quickly. “An example is the way the customer interacts with you, be that shopping or search or payments,” I was told. “Agentic Commerce is still embryonic, but the pace is enormous and we will see huge changes in this space.”
“You will start seeing agents actually working across systems and doing things reliably and autonomously, that’s when it starts to get real.”
An example of this is Amazon’s ‘Buy for Me’ experience which launched last year. The feature has been designed to use agentic AI to allow users to purchase items that aren’t sold on Amazon in the Amazon app. It’s as simple as selecting a size, clicking buy and the purchase is complete, without ever having to leave the app. This introduction of agents into the shopping experience is moving with the consumer, as 44% of users who have tried AI-powered search said it has become their ‘primary’ or ‘preferred’ source for internet searching, while McKinsey has projected that the global B2C retail market could see anywhere between $3 trillion and $5 trillion in orchestrated revenue from agentic commerce by 2030.
Behind the visible changes in search and shopping, much of the activity sits inside the organisation. I was told: “Reimagining your processes is probably 70–80% of agentic. As the tech matures even further, the “limitation” will be the ability to recognise the opportunity and apply a “revolution” mindset vs an “evolution/traditional” mindset.” And this appears to be the case, as McKinsey estimates that around 23% of organisations have begun scaling agent-based tools in at least one part of the business, a level of activity that suggests something more established is beginning to form.
In practice this often appears in areas where decisions depend on assembling large volumes of information before work can begin, similar to the way those individual users on OpenClaw were using the agents to manage their admin tasks. Buyers preparing seasonal ranges, planners reviewing stock positions or commercial teams assessing promotions will be able to arrive with much of the groundwork already assembled, changing how quickly decisions can move once discussions begin.
However, despite the opportunity that agentic AI provides, there are regulations and trust implications that organisations will need to bear in mind as it becomes embedded into everyday use. Once this AI moves from a personal tool to something businesses depend on, regulation starts to become more complicated. “The expansion of AI adoption from a personal tool to a business-critical enabler brings great opportunity but also great responsibility. All the good governance that applies to traditional applications needs to be replicated in managing cyber, data, security, identity and similar issues,” I am told. This is where the trust gap becomes critical. “If you’re automating a process where you’re trying to take the human out of the loop completely, then you’ll need to be on the money.” With only 46% of people globally willing to trust AI systems according to KPMG, organisations will need to place significant attention on getting this right, especially when taking the human out of the loop completely.
“If usage starts at the top and works its way down, it will drive companies to a culture of agent enabled change.”
All of this is new territory for senior leaders. Agentic tools introduce questions about ownership and oversight that do not always sit neatly within existing responsibilities, particularly while the work remains distributed across teams. Many executives find themselves guiding an area where experience accumulates only gradually, relying on judgement without precedent about where to proceed and where to hold back while the technology settles into place. I am told that how leadership integrates these new systems could be based on a ‘lead by example’ mantra: “If usage starts at the top and works its way down, it will drive companies to a culture of agent enabled change.
“Whether these are “big bets” and democratised smaller bets or personal productivity the momentum starts at the top, actively supported by a core technology enablement of the relevant identity, security, observability and release disciplines.”
OpenAI’s clear move towards a more agentic future is a signal for what the future could hold. As customers, we can already see how different forms of AI is entering our lives, with ChatGPT, Gemini and CoPilot already becoming irreplaceable, generally paid-for tools for many that are revolutionising both the workplace and the way individuals interact with information online. But this idea of AI as an ‘agent’ which can autonomously perform tasks doesn’t seem to have broken into people’s realities just yet, despite clear signs that they may be coming to our screens sooner than we think. What impact do you think agentic AI will have on the consumer sectors? And how are you seeing organisations adopt this new technology? I’d love to hear your thoughts.
