Agentic Commerce is fast becoming a key topic for merchants, but many are not sure where to start in evaluating how it might impact their business and understand the threats and opportunities.
This short blog helps demystify what Agentic Commerce is at a high level, the key risks to consider and what you should be thinking about before you decide your next move.
Evolution of Shopping? Agentic Commerce takes us from traditional ecommerce with a consumer interacting on a merchant’s website, to a future where a consumer’s agent (such as ChatGPT, Gemini or CoPilot and others), responds to a purchasing request and interacts with numerous merchant online services to determine the most suitable product and then complete the purchase automatically. The chances are that the consumer will never leave the Agent interface and is highly unlikely to visit your website.
On the way to this future state, there may be stages where an agent may offer purchase options for the customer to decide upon, rather than a fully automated purchase. However, as the consumer increasingly uses agents, feedback loops will improve results and learn from the consumer’s preferences reducing interaction in the process.
Merchant Site Disintermediation: The traditional merchant website will be supplemented by Agent-specific APIs that provide all necessary product data (e.g., pictures, costs, payment methods) directly to the shopping agent.
Data interfaces: are moving from web pages designed for humans to machine readable APIs and open standards—such as Google’s Universal Commerce Protocol and Model Context Protocol—which allow agents to directly access product information, pricing, availability, and fulfilment options without scraping websites.
Payments: are adapting through new agent ready rails like Mastercard’s Agent Payments Protocol and agent driven checkout experiences such as Copilot Checkout, leveraging tokenisation and ensuring that agents can complete transactions securely while merchants stay “merchant of record.”
Security: is being redesigned to handle risks unique to autonomous agents. This has led to stronger credential tokenisation, verifiable agent identity, and fraud detection systems built for agent to agent interactions.
Data interfaces: are moving from web pages designed for humans to machine readable APIs and open standards—such as Google’s Universal Commerce Protocol and Model Context Protocol—which allow agents to directly access product information, pricing, availability, and fulfilment options without scraping websites.
Payments: are adapting through new agent ready rails like Mastercard’s Agent Payments Protocol and agent driven checkout experiences such as Copilot Checkout, leveraging tokenisation and ensuring that agents can complete transactions securely while merchants stay “merchant of record.”
Security: is being redesigned to handle risks unique to autonomous agents. This has led to stronger credential tokenisation, verifiable agent identity, and fraud detection systems built for agent to agent interactions.
Merchant and Brand Erosion: The shift of control to the agent causes disintermediation of the merchant's website, disruption of meticulously designed customer journey’s leading to a loss of influence over payment method choice, and a potentially severe reduction in brand loyalty. If Search is how you, get the majority of your sales, but Search is replaced by AI search/Agents then if you're not easily visible, what happens to your sales?
Novel Fraud and Security Risks: There's a high potential for increased fraud due to new attack vectors like prompt injection, malicious agents, and the loss of traditional identity and authentication signals. This requires fraud tooling built for the AI age, not legacy products to keep pace.
Liability and Consent Ambiguity: The agent's autonomous decision-making creates significant legal and operational issues, including a lack of clear provable consent for the purchase/payment, unclear acceptance of T&Cs, and increased potential for disputes and returns where agent errors occur.
Regulatory Challenge: This model represents a fundamental shift from today's human-interaction-based payments, creating uncertainty around its compliance with current regulations like PSD2/PSR 2017 and the need for Strong Customer Authentication by the Payment User.
Optimised Conversion and Sales: By removing the human-browser friction, agents can deliver significantly lower friction payments and checkout experiences, which directly leads to higher conversion rates and increased overall sales.
Enhanced Customer Lifetime Value: For repeat purchases (like groceries or subscriptions), the reduced effort driven by autonomous agents boosts customer retention and loyalty, dramatically increasing the lifetime value of those customers.
Customer Base Expansion: Merchants who correctly structure their data, implement the necessary APIs and agent interfaces are positioned to capture a new, agent-driven customer base that bypasses traditional web search entirely.
Is this a passing fad or truly the next big thing in ecommerce that merchants must be able to support? Is it a losing battle, or should you block the bots?
Agentic Commerce will not replace traditional ecommerce overnight. Early adoption is expected in commodity and repeat‑purchase categories before moving to high‑consideration products. It has clear momentum with significant investment from large players in Big Tech and Payments.
Based on your product and demographic mix and your brand's ability to generate website traffic, when do you need to move to take advantage and not be left behind?
Suggested starting actions:
CBS Change Partners have 20 years’ experience in the changing payments landscape and are well placed to help you on the Agentic Commerce journey. Reach out to see how we can help you realise the opportunities this presents and mitigate the risks.