The promise of agentic AI
As I've previously written, over the past 12 months there has been a rising hurricane of hype around the concept of agentic AI.
Agentic AI refers to AI models that have the freedom to autonomously handle tasks. It doesn't take a great deal of imagination to picture the applications to which these agents could be put to use.
However, to date - we've not seen the promised explosion in agentic AI use amongst consumers or businesses.
Google's announcement is highly likely to change this situation.
The new Agent Payments Protocol (AP2)
As is my wont, I conducted an informal straw-poll here at Velstar towers the other day.
I asked the Owned Media team if they would entrust an agentic AI to make a purchase for them; input a detailed prompt, provide your card details, and allow the agent to complete the purchase autonomously on your behalf.
Now, bear in mind that the Owned Media team are a tech-savvy lot, spanning in age from their early-20s to early-40s.
If there was going to be a group of early adopters of agentic AI, it would be this lot.
But, to a person, they answered no. They would not entrust an AI agent to research and complete a purchase for them.
Asked why, it comes down to an issue of trust and uncertainty. What if the agent purchases the wrong item? What if the agent is defrauded by a scam website? What if the retailer's website refuses to sell to the agent?
There are a whole bevy of uncertainties and questions around agentic commerce - which, if left unanswered - will strangle the potential of agentic AI in its cradle.
Thus, Google has announced the new Agent Payments Protocol (AP2).
In Google's own words, it is 'an open protocol developed with leading payments and technology companies to securely initiate and transact agent-led payments across platforms… In concert with industry rules and standards, it establishes a payment-agnostic framework for users, merchants, and payments providers to transact with confidence across all types of payment methods'.
The AP2 has been developed precisely because today's payment systems are built on the assumption that a human is directly clicking 'buy' on a trusted surface. Autonomous agents upend this assumption and require a re-think regarding payment systems and gateways.
More specifically, AP2 seeks to address and facilitate:
- Authorisation: proving that a user gave an agent the specific authority to make a particular purchase.
- Authenticity: enabling a merchant to be sure that an agent's request accurately reflects the user's true intent.
- Accountability: determining accountability if a fraudulent or incorrect transaction occurs.
It's clear, that these are the fundamental barriers that are currently preventing the widespread adoption of agentic shopping by consumers and businesses.
Remove these barriers - as AP2 promises to do - and we could see the world of e-commerce completely changed forever.
So, how will it work?
Mandates and verifiable credentials
Google's AP2 builds trust via what it calls 'mandates' - tamper-proof, cryptographically-signed digital contracts that serve as verifiable proof of a user's instructions.
In its announcement of the protocol, Google provided the two ways in which users will shop with an agent and how mandates will work in both instances:
- Real-time purchases (human present): when you ask an agent, 'find me new white running shoes', your request is captured in an initial Intent Mandate. This provides the auditable context for the entire interaction in a transaction process. After the agent presents a cart with the shoes you want, your approval signs a Cart Mandate. This is a critical step that creates a secure, unchangeable record of the exact items and price, ensuring what you see is what you pay for.
- Delegated tasks (human not present): when you delegate a task like, 'buy concert tickets the moment they go on sale', you sign a detailed Intent Mandate upfront. This mandate specifies the rules of engagement - price limits, timing, and other conditions. It serves as verifiable, pre-authorised proof that can allow the agent to automatically generate a Cart Mandate on your behalf once your precise conditions are met.
In either case, a non-repudiable audit trail is created that provides confidence to consumers, retailers, and payment processors.
That the AP2 looks to be the future protocol for agentic transactions is backed up by the partners already committed to support the development: Mastercard, Starling, American Express, worldpay, PayPal, Revolut and more.
Imaging potentialities for agentic e-commerce
Whilst one immediately thinks of day-to-day consumers as being the 'go-to' use case for agents for shopping, the reality is that B2B applications are likely to be the biggest application of agentic AI (not in terms of absolute numbers, but in aggregate adoption).
Why?
Firstly, let's consider the direct-to-consumer e-commerce space (websites selling clothing, ornaments, gadgets etc). People enjoy the act of shopping. Part of the pleasure is in discovery; visiting various websites, perusing pictures and descriptions, mulling over a purchase and comparing to equivalents.
Consumer don't necessarily want to automate this away with an agent - the pleasure of the purchase is in the pursuit.
Secondly, let's consider the business-to-business e-commerce space (websites selling parts or spares to car garages, for example). Imagine if you're the owner of a car repair garage; putting in that weekly order for spark plugs, gaskets, air and oil filters - it's a time-consuming, boring job - taking up time that could be better spent elsewhere. This is an ideal use case for agentic AI.
We can perhaps boil the distinction down to this:
"Agentic AI will be primarily used (by consumers or businesses) to purchase mundane, consumable items".
This reality doesn't perhaps live up to the hype generated by Silicon Valley tech supremos - but it is still profoundly transformative for e-commerce brands. If you are such a brand, you may quickly find yourself retailing to both humans and agents.
So, what are the consequences of this?
Agentic e-commerce and consequences for brands
Be you a DTC or B2B e-commerce brand, the rise of agentic shopping (as facilitated by Google's new Agent Payments Protocol) will result in some major changes - to your marketing funnel, site structure and more.
Here's are my top predictions as to how agentic AI will impact your e-commerce brand:
- Intermediation jumps from search engines to shopping agents: expect less discovery and more 'agent referrals', with zero-click/summary experiences reducing site traffic.
- Price, availability, and service terms beat brand storytelling at the point of decision: agents auto-compare across retailers, track drops, and can complete purchases under user rules.
- Structured data becomes the 'new shelf': agents will rely on feeds/data (attributes, GTINs, reviews, policies). If this data is missing, thin, or inconsistent on your site, you'll be invisible to agents.
- Ops and policies need to be machine-readable: returns, warranties, and payment authorisation must be encoded, so agents can interpret them and act.
These are just a few of the likely impacts of agentic AI on e-commerce brands. Here at Velstar we'll be staying abreast of these developments and regularly reporting on them. If you want to stay in the loop, then sign up to our newsletter now.
Prepare for the future of e-commerce with Velstar
If you want to ensure your e-commerce brand is fit for the future, and gains that all important first mover advantage in the age of AI - speak to Velstar now, the home of agentic-ready websites.