Host: Andrew Birmingham, Editor - CX | Martech | Ecom
Agentic commerce is about to redraw the shopping journey. AI is moving beyond search and recommendation to evaluate products, weigh trade-offs and increasingly make decisions on the customer’s behalf.
For brands, a chatbot and a few tweaks for AI search will not cut it. Winning will require trusted data, credible evidence and enough brand strength to ensure the machines do not reduce everything to price.
In the latest edition of the Mi3 Market Voice podcast, Accenture Song Senior Managing Director and Global Commerce Practice Lead, Rajat Agarwal, joins Mi3 technology editor Andrew Birmingham to explain why AI agents could transform consumer behaviour faster than the original shift to e-commerce.
Digital commerce solved the problem of access, then replaced it with the problem of abundance. Consumers can find almost anything, but they are often left wading through endless options, dubious claims and reviews they do not entirely trust before abandoning the purchase anyway.
Agarwal says agents could break that cycle by understanding what the customer is actually trying to achieve, testing the available evidence and cutting the field down to a smaller set of credible choices.
The most immediate change will be a move from search-and-browse journeys towards intent-based commerce. Fully autonomous transactions remain at an early stage, but Agarwal expects customers to give agents greater authority as their confidence increases.
That development creates a sharper divide between brands with genuine emotional connections and those that rely primarily on broad positioning. Consumers may continue to ask agents for favoured brands, but weaker brands will increasingly be assessed through price, promotion, availability, features and supporting evidence.
“Mediocrity will evaporate,” says Agarwal who argues that brands without a strong connection to customers risk becoming commodities selected by agents according to functional attributes. “However, brands which still have the right emotional connection with the customer will continue to thrive and to do well.”
This episode also explains why agentic commerce cannot be left to a specific functional or local team or treated as another optimisation programme. Agents may draw on product pages, ratings, reviews, community discussion, video, pricing, inventory and other internal and external sources. Brands need consistent, structured and credible information across that entire environment.
Agarwal advises companies to begin with the customer journeys they want to win. They should then identify the evidence an agent will need, create the supporting knowledge graph, improve product-information and digital-asset systems, distribute the information consistently and measure whether the strategy is working.
Starting with a generic AI-visibility score can produce the wrong result. Agarwal cites a fashion business that appeared to perform strongly in agent discovery but was being associated with discounts and value rather than fashion, inspiration and style.
Choice of agents, agents of choice
The episode develops an important distinction between becoming a choice of agents and becoming an agent of choice.
Every consumer-facing business will need to become a choice of agents by ensuring its products can be found, understood and purchased through external AI systems. Agarwal calls this a no-regret move. “You have to make your product discoverable and transactable by these horizontal agents.”
A smaller group of companies will aim higher. Rather than simply making their products visible to other people’s agents, they will try to build the agent customers choose first. Done well, those vertical agents could combine deep category expertise and human judgment with the speed and convenience of digital commerce.
The prize is establishing and preserving intimate customer relationships.The risk is that the agent carries the brand with it. If it hallucinates, makes a poor recommendation or loses the customer’s trust, the damage will not stop with the technology.
Agarwal’s advice is to start now, without pretending the data, models or architecture are finished. Agentic commerce needs to be built as an enterprise capability spanning marketing, commerce, communications, supply chain, technology, governance and the operating model.
The business case runs well beyond cost cutting. Agentic commerce could lift conversion, rescue sales lost to choice overload, reduce acquisition costs and give brands more room to move without adding spending at the same rate.
Agarwal tells Mi3 that the early results are promising. The catch is that the market is still too young, and the evidence too uneven, to produce reliable benchmarks that every company can plug into a business case.
His 12-month prescription is straightforward. First, understand how the brand is being read across the AI decision ecosystem. Then fix the data and technology foundations. In parallel, companies should decide whether their own websites, apps and commerce platforms need to become agentic experiences.
The winners will be the companies that treat agentic commerce as a fundamental change to how they operate. The rest risk accumulating pilots, presentations and proofs of concept that look impressive in the boardroom but amount to little more than AI theatre.
See omnystudio.com/listener for privacy information.