The tech and financial industries are excited about AI agents that shop on behalf of consumers, but a product manager at Hedera AI Studio says that vision is not functional yet. Mastercard's report projects agentic commerce could reach $1 trillion in US consumer retail by 2030, yet the source notes that "could" is doing a lot of work. Reality checks are already piling up: Amazon blocked Meta's Muse bot and previously blocked Perplexity's AI bot, citing terms of use restrictions.
Lindsay Walker, product manager at Hedera AI Studio, argues that agentic commerce spans discovery, negotiation, settlement, and post-purchase logistics, not just the payment moment. Simple AI-assisted product selection is viable, she says, but things break down when users delegate authority to an agent that makes decisions on its own. Money is too important to be left to a probabilistic model, so transactions need deterministic gates that cannot be bypassed.
There are also basic functional problems: Walker tried to use an agentic wallet or card on a website, but the AI agent could not log in to access her account or fill in the credit card field on the page. She contends the industry needs a mechanism tethering a person's identity to an agent. The Register's own analysis adds skepticism, pointing out that e-commerce already offers automated repeat purchasing, low-cost items are trivially easy to buy, and high-cost purchases demand human care. Trust and regulatory scrutiny of personalized pricing further complicate the path, so the source concludes that agentic commerce is years away.