Karsten Wenzlaff, Advisor
August 26th, 2025
Karsten Wenzlaff is a Crowdfunding and Fintech Expert from Germany. He is Secretary-General of Digit[...]
Should Fintechs Design for People or AI Agents?
Agentic AI | May 5, 2025
Image: Freepik
Visa and Mastercard Join the AI 'Buy for me' Race. What Should Fintechs Do Now?
Last month, Amazon launched its AI concierge shopping service. Now Visa is rolling out its 'Intelligent Commerce' (April 30) and Mastercard with its 'Agent Pay' (Apr 29). Without question artificial intelligent smart, buying assistants are becoming active customers in daily transactions. They aren't just answering questions or offering a comparative analysis, they are being enabled to make real purchases, book real services, and handle real payments. While these services are being tested and iterated, the implications are significant for fintech companies. If people rely on smart agents as (one of) the main ways they shop and pay, then financial technology firms that aren't ready could be left behind.
Are fintechs still designing for people?
Agentic AI is flipping the script and changing the role of the actual human buyer. More and more decisions and purchases on behalf of customers will be completed by agentic AI buyers, meaning that the front-end UX/UI will also need to be designed for AI agents, and overtime less about human buyers. So traditional fintech interfaces that rely on pleasing human designs, well organized navigation, step-by-step forms and processes, may soon be irrelevant.
See: Why No Code AI Agents Matter for Fintech in Canada
Why? Well smart agents don't browse; they query APIs based on a set of rule-based purchasing requirements, and act instantly. Fintechs now need to build tools that interact with agents and not just design interfaces that are pleasing for humans to look at. As the sentence is written, starting to wonder if we're all turning into robots here!
What are "buy-for-me" models?
Instead of a human user buying something directly, they authorize their agentic AI buying assistant to buy it for them. These 'buy-for-me' agents follow rules set by the end-user, such as only buy from my trusted list of sellers, or stay within a budget. Important to know that these agents don't use traditional credit cards to make a purchase, they are using safer, digital payment tokens. Visa and Mastercard have already built the infrastructure to enable agents to act on behalf of users (no credit card required).
What are the biggest challenges?
At the core of so many digital innovations, and perhaps one of the biggest challenges with Agentic AI buyers, is 'TRUST'. People are still unsure whether to let an AI spend their hard earned money, especially for large purchases like a new car, TV or even a house. Even with the right rules and controls in place, users need a way to monitor or check in on what an assistant is doing. If an agent buys something on their behalf, they may for example expect to have the ability to override or undue mistakes quickly, so fintechs must offer tools that make this experience easier and more seamless than today, if they want adoption to grow.
See: Balancing AI Automation and Ethics in Fintech
Then there's regulation. If an agent hallucinates after explicitly being instructed to follow a laundry list of rules, then they purchase something that's not wanted or more expensive than allowed then who is actually responsible? Right now the rules are unclear. So financial tech firms will need to show a traceable, non-tamperable, audit trail to prove that the agent had permission to buy (or not). Financial service transactions are highly sensitive and private, rules and the regulations around them must be strictly enforced, or risk backlash.
Another challenge will be interoperability between different systems. Amazon, Visa, Mastercard, and others are each building their own platforms. Without shared standards, fintechs may have to build separate connections for each system. Creating a shared standard that optimizes interoperability is a great opportunity and challenge for facilitating how AI agent-driven commerce could be globally.
Can AI agents already shop and buy today?
Yes, within certain limits. Visa lets users set conditions for how their assistants can spend. Mastercard supports secure payments using special tokens built for AI. These buying systems are live and consumers can already use them today but most buying tools are still restricted to basic purchasing functionality, and can't yet handle more complex decisions like comparing options or negotiating pricing deals and discounts.
Will we see general purpose or niche agents?
AI agents are still an emerging innovation, so we'll likely see niche agents for a while, such as one to help buy groceries, or find discounted prices, book travel, or manage recurring business purchases. General AI buying assistants or ones that can conceivably buy anything from anywhere are still many steps away. They'll need access to more interfaces, more data, and optimized logic to support complex decisions and outcomes.
What should fintechs do right now?
Now is the time to prepare and build digital wallets that work well with AI buying agents. These smart wallets should show what the assistant is doing and let users set rules or stop transactions if needed. Fintechs should also think about offering services that agents can call on. For example, an assistant might ask for the best foreign exchange rate or check for promotional offers. Fintechs that provide this kind of useful data to agents will be in demand.
See: How Real-Time Agentic AI Will Boost Fintechs
Another need is for smarter fraud detection. Agentic spending looks different from human spending, as it's faster and sometimes harder to explain. This means that older fraud detection models won’t work as well. Fintechs that build tools to understand and respond to this new kind of buying activity will have an edge.
Consent and approval tracking is another piece of the puzzle. If an agent buys something, there needs to be a clear record showing what it was allowed to do and when. Building this kind of audit trail will be essential, as regulators and buyers will demand it.
Closing Outlook
As financial and e-commerce giants like Visa, Amazon, and Mastercard rollout smart buying agents, fintechs need to start preparing to build wallets and systems that work with these agents. To offer services and data in a way that AI agents can easily use. Tools must be safe, compliant, flexible, and transparent. Early companies that build this changing AI agent buying layer will be ready for what agent marketplaces will look like and how they behave in the future. Not to long from now, human users will be choosing which agents, the marketplace, and fintech wallets they want handling their purchases and money.
The National Crowdfunding & Fintech Association (NCFA Canada) is a financial innovation ecosystem that provides education, market intelligence, industry stewardship, networking and funding opportunities and services to thousands of community members and works closely with industry, government, partners and affiliates to create a vibrant and innovative fintech and funding industry in Canada. Decentralized and distributed, NCFA is engaged with global stakeholders and helps incubate projects and investment in fintech, alternative finance, crowdfunding, peer-to-peer finance, payments, digital assets and tokens, artificial intelligence, blockchain, cryptocurrency, regtech, and insurtech sectors. Join Canada's Fintech & Funding Community today FREE! Or become a contributing member and get perks. For more information, please visit: www.ncfacanada.org
 |  |  |
|---|---|---|
 |  |  |
Leave a Reply