What Is Agentic Commerce? Hype, reality, and what it means for fintech

It got me thinking:

• Will Agentic Commerce be a monumental paradigm shift in the human experience as many people are purporting it to be, or simply an iteration on what we are already doing?
• Will Agentic Commerce live up to the hype or be consigned to the dustbin of history like other buzz phrases such as Internet of Things, Metaverse and Blockchain?
• Or will it have staying power, driven by real-world application?

What is Agentic Commerce?

‘Agentic’ is an awkward-sounding term. We should get comfortable with it, as the hype engine around AC is just getting started.

AC involves AI programs that help a buyer to complete their purchase. Agentic = AI agents and bots. Commerce = buying and selling things.

In AC, the theory goes that the buyer – an individual or a business – completes a purchase using an AI program or another tool – such as a search engine – that supports such functionality.

In future, common AI platforms such as ChatGPT, Gemini et al may support AC purchase experiences when shopping online, via integrations with our browsers and apps, on computers and mobile devices.

This could include product search, product display, checkout, payment and order confirmation steps. The AI program will dispatch a bot to complete the purchase while the buyer is present in the moment, or while the buyer is absent from the moment after having set-up the process to automate it.

Why is Agentic Commerce getting such attention now?

Broadly, the huge uptick in discussion of this topic has been driven by AI products that have sprung up recently, such as OpenAI’s ChatGPT, Google’s Gemini, Microsoft’s Copilot and Anthropic’s Claude. To name but a few. The pace at which mass adoption of AI tools has occurred is dizzying.

AI is now in our homes, on our desks and on the agenda in boardrooms globally to an extent that was unimaginable just a few years ago.

Through the fog of marketing, gossip and hype, real-world practical applications of these technologies are beginning to crystallize.

Agentic Commerce is an example of a real-world use case with almost universal application: shopping.

What impact is Agentic Commerce having on payments and fintech companies?

Big players

We are already seeing the industry respond by announcing new AC products and adaptations to existing services.

Visa and Mastercard have recently announced their respective services which cater to the AC use case:

• Visa Intelligent Commerce: https://corporate.visa.com/en/products/intelligent-commerce.html
• Mastercard Agent Pay: https://www.mastercard.com/news/press/2025/april/mastercard-unveils-agent-pay-pioneering-agentic-payments-technology-to-power-commerce-in-the-age-of-ai/
• American Express has invested in agentic commerce start-ups, including Nekuda: https://www.businesswire.com/news/home/20250514808097/en/Nekuda-Raises-%245M-Led-by-Madrona-Together-with-Amex-Ventures-and-Visa-Ventures-to-Power-Agentic-Payments

Smaller players

Start-up companies are piling on to this trend. They are coming into the market with AI and AC products that purport to solve for relevant uses cases. This rise of new entrants will soon reach saturation point. Many will fail. Some will merge. Larger players will buy the best and most promising of the smaller players.

Billions of dollars in PE and VC investment will change hands, much of it to never be seen again.

A few names will achieve ‘escape velocity’ and survive long enough to enter the popular consciousness as sub-brands and tools of larger organisations.

What does Agentic Commerce mean for individuals and businesses?

Many businesses and individuals working in the industry will brand themselves as Agentic Commerce experts. However, we are all at the starting line of this technology and learning about it as we go. So, let’s act with proportionate caution and be kind to ourselves as our understanding evolves.

I could not claim to be an expert in Agentic Commerce or AI, but I’ve been in the world of payments and fintech long enough to be able to form instinctive reactions about such topics. Here are some of my hypotheses with regard to AC:

• Human behaviour around shopping is deeply entrenched. Many people enjoy shopping, and for some it is a big part of their lives. This is not going to change because of AI-powered shopping assistants.
• More likely is that we will continue to split our shopping and purchasing behaviour between a variety of channels and tools. Increasingly, AC will form a certain percentage of that activity. But not replace it wholly. The future of e-commerce is more options, fewer concentrations.
• We will continue to browse apps, websites and marketplaces and search for products in the same way we do today – because we want to make more careful decisions about what we buy or want to enjoy the experience of shopping. Sometimes we will choose to use an AC tool to make the process a bit quicker and more efficient. Perhaps we may care a bit less about the outcome and want to speed things-up a bit or wish to have a more curated set of options to review.
• Gradually, people will start to use AI programs and AC tools to help them make everyday purchases. Consumer goods and services such as groceries, electronics, fashion and bill payments are likely to be the first movers. Other sectors will follow.
• Businesses that sell D2C or B2B/2C will still operate as they do today but will need to adapt purchase channels and flows to AC behaviour, including bots. The starting point for a shopper will remain their laptop, computer or device, but AC software will necessitate businesses to iterate on their existing technical protocols, data flows, authentication, risk & security frameworks in order to be made sensitive to and compatible with AC tools.

Is Agentic Commerce a revolution or an iteration?

My theory is that AC is simply an iteration on existing processes and behaviours.

One challenge is bots. We’ve spent the past couple of decades in e-commerce attempting to fight bots, because most bot activity is malicious or fraudulent. The industry will need to adapt to new types of bots acting at the behest of AC programs.

Part of the solution to sorting the wheat from the chaff here already exists in the form of tokens. Credit and debit card tokenisation has been around for a while now and has gained – and continues to gain – widespread adoption by stakeholders across the value chain. A token can replace the full card number when the shopper enrols in a digital wallet or browser widget, and these tokens can act as a strong trust indicator, signalling that an AC bot is acting with the buyer’s authority. Authentication is completed at the ‘front end’ AC tool, embedded in the app/site/search tool – similar to (and likely using) digital wallets such as Apple Pay, Google Pay and PayPal.

Tokens, token meta data and authentication flags will help reduce friction and false declines due to fraud prevention and detection software used across the AC-originated payment lifecycle – in particular on the merchant side.

E-commerce websites and apps, payment gateways and acquirers, fraud detection software and tools, card schemes and card issuers can all utilise existing systems, processes and data flows to enable AC purchases with fairly minor adaptations and nuances.

So, is Agentic Commerce a revolution or an iteration? Right now, it looks like the latter — but history has shown us that small iterations can become major shifts. How will your business prepare?