A new category of generative AI, known as agentic AI, is emerging, sparking intense competition among specialists such as OpenAI, Perplexity, Anthropic, and Mistral AI, as well as major tech companies eager to leverage their substantial investments in artificial intelligence.
These advanced autonomous AI systems are capable of reasoning, navigating databases, software, email, and other Customer Relationship Management (CRM) systems. They can interact with other AI agents and manage complex tasks autonomously.
In the realm of agentic commerce, they can function as personal AI shoppers, offering an integrated shopping experience. Based on a buying intention articulated in natural language, the AI agent can manage the entire purchasing journey - searching for products, comparing options, providing recommendations, and even validating the shopping basket and completing payment on behalf of the user.
In her latest report, Micaella Feldstein, Tech & Data Specialist, discusses how this technology might reshape e-commerce, especially if it successfully addresses critical obstacles.
The time savings, personalized shopping experience, and ease of use of agentic commerce are likely to attract consumers. However, fully delegating the purchasing process, particularly the payment stage, poses a challenge for now, despite advances in security technology.
Micaella Feldstein
Implications for the E-Commerce Ecosystem
While these capabilities are still in early development, the shift toward agentic commerce will undoubtedly impact various stakeholders within the retail ecosystem. According to McKinsey, AI could generate an additional $240 to $390 billion in annual e-commerce revenues. The potential disruption to the dominance of giants like Google, Microsoft, and Amazon is significant, and these companies are already responding by investing in agentic AI.
This new form of commerce will prompt brands to enhance their online visibility through innovative strategies such as Generative Engine Optimization (GEO), which will replace traditional Search Engine Optimization (SEO). User interfaces are likely to include voice AI, visual search, virtual fittings, and expedited delivery services - all aimed at creating an experience akin to that of brick-and-mortar stores.
However, reluctance among retailers concerning the resulting disintermediation may hinder the deployment of agentic systems, as it poses a serious threat to customer relations particularly through the loss of first-party or proprietary data and targeted advertising. In this model, retailers may transition to mere data providers, offering information on stock levels, prices, and product specifications.
Consumer Trust and Operational Security
Moreover, the acceptance of agentic commerce critically depends on consumer trust. Transparency regarding how agentic AI operates, particularly concerning sponsored recommendations, is essential. As AI agents potentially evolve and interact with vast amounts of data - including payment information, personal details, and operational metrics - they introduce new cybersecurity risks, necessitating comprehensive strategies to protect user data and secure transactions.
Notably, enabling AI to process payments requires the development of a robust payment architecture. Major players in the payment industry, such as Visa, Mastercard, and PayPal, have already launched initiatives to adapt their payment infrastructures to the specificities of agentic commerce. These initiatives emphasize tokenization, user-defined controls, and seamless integration into conversational AI interfaces to establish the trust necessary for widespread adoption.
Towards a Hybrid Adoption
Agentic commerce is still in its infancy, and it is likely to be initially embraced by consumers for product research, discovery, and comparison - similar to the rapid adoption of digital assistants like ChatGPT.
However, fully delegating the entire purchasing process to AI, especially during payment, will be a more challenging leap despite advancements in security technology. We foresee that complete delegation will likely be trialed by consumers for routine, low-cost, and uncomplicated purchases, such as groceries and cleaning products.
