Agentic commerce promises frictionless consumption — AI systems that search, decide, and transact on our behalf. However, the removal of human action at the moment of purchase erodes the central premise of current payment infrastructure: cardholder intent. Agentic commerce fails because it breaks the alignment between intent, execution, and accountability. So far, we hold no solutions.
Today’s commerce stack builds on the assumption that a human has authorised a transaction. Every safeguard downstream, from fraud detection to chargebacks, functions to reconstruct that intent when something goes wrong. The responsibility, or risk, is authorised by the agent, or buyer. AI agents introduced delegated intent without verifiable context.
We see a layer of abstraction introduced through agentic commerce that current payment and dispute systems cannot verify, making trust harder to keep.
“This is where AI and machine learning become essential on the resolution side, not just the transaction side. Chargebacks911 addresses this through its Unified Dispute Management System (UDMS) and ResolveLab, which use AI and machine learning to analyse dispute data, improve operational visibility, and support faster, more informed decision-making across global markets.”
“Visa is building the front end of agentic commerce, but the dispute infrastructure will need to evolve just as quickly.”
Monica Eaton, Founder and CEO of Chargebacks911
ERIS SAYS: Until systems can interpret intent with reliable accuracy, after-the-fact trust will remain a bottleneck. Miranda Eaton touches on the missing key — explainability — suggesting that a move beyond clear parameters and liability is necessary. Agentic commerce technology must evolve past decision-making and focus on decision justification for sustained and effective use. Without a justification layer, trust will continue to fade and any attempts at functional AI consumption will remain futile.