Ecommpay report key takeaways – trust and control in agentic commerce

AI agents are rapidly moving from theory to practice, evolving from tools that assist commerce to systems capable of acting on their behalf. Industry forecasts suggest that AI agents could handle up to USD 5 trillion in global transaction value by 2030, promising better convenience, automation, and efficiency across the shopping journey. 

However, according to Ecommpay’s latest report, Before they check out: Why payment service providers should deliver trust in agentic commerce for merchants and customers, the technology may be moving faster than customer trust. 

Trust, but verify, as former US President Ronald Reagan famously put it, becomes increasingly relevant in the context of agentic commerce, where decision-making is gradually shifting from humans to autonomous systems. 

Based on a survey of more than 1,750 consumers across the UK, France, Germany, Spain, and Italy, alongside interviews with merchants and payment industry stakeholders, the report examines what is needed before agentic commerce can scale.

Here are some of the report’s key takeaways. 

Consumers trust AI for discovery, not for checkout 

The research reveals a clear distinction between how consumers use AI today and how willing they are to let it make purchasing decisions. 

While 55.9% of respondents have already used AI tools for search-related activities and 60.1% say they are comfortable with AI in general, enthusiasm declines when payments enter the picture. 

Consumers see the greatest value in practical shopping assistance, including comparing prices (53.3%), finding discounts (47.1%), and recommending products based on personal preferences (41.1%). However, only 13.9% would trust an AI agent to complete a purchase on their behalf. 

The findings suggest that consumers currently view AI agents as research and recommendation tools rather than autonomous buyers. 

Control – a central key to consumer trust

One of the clearest messages throughout the report is that trust depends on maintaining consumer control. 

More than half of responders (50.1%) would not trust an AI agent with their card details, while 44.3% would limit agent spending to EUR 100 or less. Only 6% would allow an agent to spend more than EUR 500. 

Consumers identified several safeguards that would make them more comfortable with agentic commerce, including approval before purchases, spending limits, additional verification steps, and straightforward refund processes. 

The findings suggest that successful agentic commerce adoption models will likely rely on controlled delegation, giving AI agents permission to act within clearly defined limits while maintaining transparency and consumer oversight throughout the purchasing process.

Fraud, liability, and accountability remain unresolved

The report highlights a growing gap between existing ecommerce infrastructure and the realities of agent-driven transactions. 

Traditional fraud prevention systems rely heavily on behavioural signals generated by human users, such as fingerprints, browsing behaviour, and typing patterns. As AI agents begin to act independently, many of these signals will become less reliable.

Consumers are also uncertain about accountability. Nearly 39% believe AI companies should be responsible for fraud-related losses, while almost 24% point to banks and payment providers. 

For merchants, questions about chargebacks, dispute resolution, liability allocation, and fraud management remain largely unanswered. 

Agentic commerce will require a phased approach 

According to the report, fully autonomous checkout is unlikely to become mainstream overnight.

Instead, the report suggests a gradual evolution, moving from AI-assisted discovery and recommendations towards bounded-authority models, where agents can act only within predefined spending limits and approval frameworks. 

This phased approach allows consumers, merchants, and regulators to build confidence while developing the operational and legal safeguards required for wider adoption. Additionally, the trust mechanisms must evolve alongside technology, rather than after deployment.

PSPs: the trust infrastructure behind agentic commerce

As AI agents become part of the payment journey, payment service providers are well-positioned to become the trust layer that enables agentic commerce to scale safely. 

PSPs already sit at the intersection of merchants, consumers, issuers, and payment networks, giving them a unique position within the ecosystem. Beyond processing transactions, they have the authentication capabilities, fraud expertise, operational experience, and network relationships needed to translate emerging agentic commerce models into secure and practical payment experiences.

According to Ecommpay’s research, merchants consistently identified PSPs as central to building the foundations of trust in agentic commerce. This includes establishing standards for authentication, tokenization, transaction mandates, fraud prevention, dispute management, and data sharing. Their experience implementing frameworks such as 3-D Secure also positions them to help shape the protocols that will govern future agent-driven transactions. 

As consumer trust declines when spending authority, payment credentials, and autonomous purchasing enter the picture, PSPs could become the infrastructure that balances automation with control. 

What’s coming next?

The report concludes that agentic commerce has the potential to transform how consumers discover, evaluate, and purchase products online. However, widespread adoption will depend less on technical capability and more on trust. 

In agentic commerce, trust may become the ultimate payment credential. Consumers appear willing to embrace AI assistance, but only when control, transparency, and accountability remain firmly in their hands.

Agentic commerce will not scale on capability alone; it will scale when consumers and merchants have the confidence, controls, and accountability they need to trust it. While AI agents have the potential to transform how the products are discovered and purchased, wider adoption will depend on whether users feel in control of the decisions being made on their behalf.

Trust becomes the defining requirement, shaped by control, transparency, and clear liability frameworks.

For merchants, PSPs, and technology companies, the challenge is no longer whether agentic commerce will emerge, but how to build the trust infrastructure required to support it.

Want to learn more? Download Ecommpay’s report, Before they check out: Why payment service providers should deliver trust in agentic commerce for merchants and customers.