The End of Clicks: What Happens When AI Agents Become Your Customers (Part 1)
Why the real future of AI isn’t chat - it’s autonomous systems negotiating commerce behind the scenes
Hi everyone, thanks for coming back to Customer Futures.
Each week I unpack the disruptive shifts around ‘Empowerment Tech’. AI Agents, digital wallets and new digital customer relationships.
If you haven’t yet signed up, why not subscribe:
Hi folks 👋
Today’s post is another special edition in the Empowerment Tech Deep Dives series.
Gartner says that ‘machine customers’, including AI agents, will be “a market shift roughly twice as large, and twice as fast, as the historical arrival of e-commerce.”
If they are right - and from what I can see happening in the market right now, I think they might be - we need to start paying attention to the combination of AI Agents and digital wallets. Specifically to ‘A-Commerce’.
Because our digital experiences are about to be turned inside out.
And I can’t think of anyone better than Gam Dias to help us make sense of it all.
He’s been working on ‘personal data control’ for over a decade. But most interestingly, he’s been looking at it from inside the business. Exploring personal data portability, process automation and customer empowerment from the perspective of CRM, customer data platforms and digital twins.
And now, most excitingly, from the point of view of Agentic AI.
Gam is walking the talk with AI. Using LLMs and agents to research and co-author a series of books on the topic. His latest, Agents Unleashed, came out last month.
Today’s post is a preview of one of the chapters in his next book, Agents Unleashed: The Autonomous Edge. Gam walks us through the impact of AI Agents in a specific sector: travel and hospitality.
It’s a long one today, where we cover:
The familiar frustration of booking travel
The risk of AI Agents (and the UX shift they trigger)
Enter ‘Model Context Protocol’ (MCP) - without the buzzwords
A day in the life of a headless travel booking
The digital wallet is coming for everyone
Why removing friction changes the game
In Part Two next week, Gam will look at how AI agents actually communicate, and the profound implications for the digital customer experience.
There’s a LOT going on right now. I’m speaking to quite a few people about how Personal AI will impact their business and customer engagement. Especially around Anthropic’s Model Context Protocol (MCP).
If you’d like to talk about any of it - from MCP and AI Agents, to the disruption coming with digital wallets and portable digital credentials - we’re here to help.
You can book a call with Gam and I here.
For now though, grab a hot brew and a comfy chair, and Let’s Go.
The familiar frustration of booking travel
You know the drill.
You need to book a hotel for a work trip or a quick getaway. You open a browser tab, maybe two, maybe twenty, and begin the ritual.
Search. Filter. Compare. Wonder if the price will go up if you wait. Wonder if the price already went up because you didn’t. Try to remember your loyalty logins. Try to recall if you stayed at that property last year or if it just feels familiar because every booking site looks the same.
It’s not that the tools are broken. Expedia, Booking.com, Hotels.com - even the hotel chains’ own sites - have all improved over time. But they were designed with a specific kind of user in mind: someone who takes the wheel.
These platforms assume the customer is active, decisive, and willing to do the work. In return, they offer choice, but not ease. And so, we’ve come to accept friction as part of the process.
The deeper issue? These systems are built for the average traveler, not for you.
Maybe you’re a light sleeper who avoids elevators. Maybe you need two beds because your child often joins. Maybe you care more about natural light than price, or you’re using points and flexibility matters more than location. Good luck capturing any of that nuance in a filter panel.
So we adapt. We open tab after tab. We scroll, compare, and second-guess. Not because we want to, but because the system requires it. It has no idea who we are.
That’s about to change.
Because behind the scenes, a quiet revolution is unfolding. Your digital assistant won’t just suggest options, it will book them. Businesses won’t just design interfaces for people, they’ll expose services for agents. Machines will talk to machines, not to replace human choice, but to reflect it with more precision.
This is the threshold of agentic commerce.
And like the first time you tapped “Buy Now” on Amazon or watched your Uber driver approach in real time, once you experience it, there’s no going back.
The risk of AI Agents (and the UX shift they trigger)
When ChatGPT first appeared, the reactions were loud and divided. Some marveled at its fluency. Others mocked its flaws. But nearly everyone overlooked the real breakthrough. This wasn’t just about generating answers. It was about something deeper: delegating intent.
The novelty of AI may have started in conversation, but the real transformation is happening in agency. What changes everything is not that software can talk, but that it can act. That it can take your goal and turn it into an outcome, without handing the work back to you.
We’ve crossed a line. Today’s AI agents don’t just reply. They remember. They reason. They execute.
Back to the travel example: imagine saying, “Book me something in Rome under $500, walking distance to the Trastevere district. Needs a balcony. And don’t show me anything with a rating below 4.2.”
An agent can already parse that request. What’s changing is what happens after.
It doesn’t just return a list of links. It gets to work. It visits booking platforms, filters results, checks availability, compares prices. And if it finds a match within your parameters, it can book it, securely, on your behalf while you’re making coffee.
That’s not a better chatbot, that’s a shift in how digital systems behave. And it’s just beginning.
The shift is subtle but seismic.
We’re moving from search-and-click, to state-your-intent-and-trust. From user interfaces to action interfaces. From websites designed for people to services designed for agents.
That changes how product teams build. It changes how businesses differentiate. It even changes what it means to own a brand. Because soon, it won’t just be humans who engage with your digital storefront. It’ll be their agents.
And if your system doesn’t understand how to talk to them, you’ll never even know the customer showed up.
This is why companies like OpenAI and Anthropic are racing to formalize standards like the Model Context Protocol (MCP). Because without structure, agents can’t negotiate. Without memory, they can’t personalize. And without guardrails, they can’t be trusted.
What started as a smarter chatbot is becoming something much bigger. A new interface layer for the internet, one built not around content, but around context.
The rise of AI agents isn’t just a technical evolution.
It’s a UX inversion, and we’re only beginning to see what that unlocks.
Enter ‘Model Context Protocol’ (MCP) - without the buzzwords
Let’s pause for a moment. If you were hiring someone to handle hotel bookings, you wouldn’t just say, “Get the customer a room.” That’s not a briefing, it’s a gamble.
Instead, you’d offer real context. Who the customer is. What they value. What systems they can use. What the budget is. Maybe even what went wrong - or right - the last time. You’d equip them with enough background to make good decisions on your behalf.
Now, AI agents need the same clarity.
So how do you give a machine that kind of clarity before it acts? That’s exactly what the Model Context Protocol (MCP) was designed to do. Think of MCP as the briefing layer between you and the agent. It doesn’t just send a request, it sends purpose. Structured, scoped, and explicit. It tells the agent:
What its task is
What tools it can access
Who it’s working for
What constraints to follow
What’s already happened in the session
And what not to touch
This isn’t prompt engineering... it’s governance by design.
In technical terms, MCP is like ‘USB-C for AI’. A standard connector. One that lets any compliant agent interface with any compliant service, without needing a custom cable for every new integration.
Early AI systems relied on guesswork, and inferred intent from loosely structured input. MCP replaces that ambiguity with alignment. It introduces predictability, enforces boundaries, and - critically - makes agents interoperable. Your personal assistant and a third-party travel site can now speak the same language, even if they were built by different vendors on entirely different stacks.
That’s not just an engineering detail - it’s now a foundation for trust.
Because in a world where agents act on your behalf, trust doesn’t come just from encryption or privacy policies alone. It comes from knowing the agent understands your intent, your permissions, and your limits. That’s what MCP makes possible, quietly, but powerfully.
It’s the difference between handing tasks to a random consultant… and delegating to a well-briefed executive who knows exactly how far they’re allowed to go.
As agents become your digital frontlines - and as commerce shifts from screens to conversations - protocols like MCP are what make the entire ecosystem safe, personal, and scalable. And just like HTTP transformed the web, MCP might quietly transform how business gets done.
When agents become your digital frontlines, and as commerce shifts from screens to conversations, protocols like MCP are what make the entire ecosystem safe, personal, and scalable.
And here’s why that matters. The last revolution in travel came when online booking platforms disintermediated travel agents, offering a seemingly infinite array of self-service options. But they left the customer with the burden of navigating that complexity alone.
Agentic experiences turn that model inside out. They restore the convenience of delegation, but with more choice and control than a human could manage alone. If travel booking can be ‘agentified’, then virtually any purchasing interaction can be.
That’s not a small shift. It’s a blueprint for how the software will serve us next.
A day in the life of a headless travel booking
It’s a Wednesday afternoon. You’re in the back of an Uber, replying to emails and trying to stay one step ahead of your calendar.
But then you remember your friend’s birthday is coming up, and you promised to fly out and visit. No panic. No last-minute scramble. You don’t reach for an app or open a travel site. Instead, you say aloud, without thinking twice:
“Book me a hotel in Austin next weekend. Under $250 a night. Central, quiet, with a king bed. Same place I stayed last time if they’ve got availability.”
Your agent responds simply with “Got it. Checking availability.” And you go back to your inbox.
Behind the scenes, a different kind of transaction is underway. Your personal agent has already reached out to several travel platforms. Each one is equipped with its own MCP-enabled AI interface. No screens, no clicks, no humans in the loop.
This is not just online commerce. It’s headless commerce: machine-to-machine interaction driven by context and executed with trust.
The hotel’s agent doesn’t receive a vague string of text. It receives a structured context bundle. Your verified identity, your prior booking history with the property, your sleep preferences, your loyalty tier, and your budget range.
It doesn’t need to guess or ask clarifying questions. It just knows. It checks for availability, applies loyalty discounts, filters out rooms near the elevator, and returns three matching options.
Your agent reviews them, weighs the trade-offs, and selects the best fit. It authorizes payment using tokenized credentials from your digital wallet, no manual entry required. A hold is placed. A receipt is stored. The itinerary is synced to your calendar. The hotel and flight are now coordinated.
You didn’t fill out a form. You didn’t browse. You didn’t compare. You simply expressed intent, and let your agent do the rest.
What might sound like science fiction is, in fact, already possible.
The building blocks are all here. Agents like OpenAI’s Operator are emerging. Standards like the Model Context Protocol allow these agents to communicate. And now digital wallets not only payment credentials, but identity and preference credentials too. AI systems are evolving from suggestion engines into execution engines.
The technology is no longer the question. The integration is. Because we’re not just digitizing commerce, we’re decoupling it from the user interface entirely.
In this world, businesses won’t compete for clicks. They’ll compete to be included to be discoverable, legible, and actionable within an agent’s consideration set.
And that changes everything.
The digital wallet is coming for everyone
A decade ago, the idea of paying for coffee with your face or fingerprint felt like sci-fi. Today, biometric authentication is so routine we barely notice it. Now imagine that same shift happening with everything else about you.
Your ID, your preferences, your credentials, your loyalty cards, even your travel habits, all stored in a digital wallet you don’t just carry, but that your AI agent uses on your behalf.
This isn’t speculative. It’s already underway.
In the European Union, the EU Digital Identity Wallet is rolling out in 2026, giving citizens a standardized way to store and present identity, health cards, diplomas, and more, securely and selectively.
And in the US, Apple Wallet and Google Wallet are quietly expanding their roles beyond payment, adding driver’s licenses, state IDs, boarding passes, transit cards, and soon, verifiable credentials issued by governments and enterprises. In India, Aadhaar already links biometrics, payments, and identity across billions.
All around the world, the digital wallet infrastructure is being built. But here’s what most businesses haven’t yet realized.
The wallet isn’t just for the consumer. It’s for the agent.
Your AI assistant will need to prove who you are. It will need to present your travel preferences. It will need to verify that your credit card is authorized for the transaction. And it will need to do all this without leaking data or asking you to re-authenticate every time it acts.
The digital wallet becomes the trust layer in agent-to-agent commerce. It carries:
Identity: Verifiable credentials signed by issuers (like a passport or loyalty status)
Preferences: Your seat choices, hotel tiers, dietary needs, budget ranges
Permissions: What your agent is allowed to do, spend, or share
Payment tokens: Secure instruments for holding, reserving, or booking
All of this can be accessed by your agent, surfaced as context, and shared with systems like travel agents via MCP, with fine-grained control.
What cookies did for personalized browsing, wallets will do for personalized execution.
So if you’re building digital experiences and not preparing for a future where agents interact through wallets, you may soon find your UX sitting empty.
Because your customer didn’t leave you. Their agent just didn’t see a reason to engage.
Why removing friction changes the game
Technology adoption doesn’t happen because people want new tech. It happens because old frictions become intolerable.
That’s what made Uber unstoppable. It wasn’t the map, the car, or the app. It was the ability to skip the awkward call, the long wait, the cash exchange. And Amazon Prime? Same story. The one-click promise wasn’t speed, it was trust that speed would deliver the right thing.
Now apply that lens to travel. Where today booking a hotel means you:
Juggle dozens of open tabs
Compare ratings that barely differ
Re-enter payment details
Recheck your calendar
Second-guess yourself, especially under pressure
This is tolerated because we’ve internalized it. But it’s not acceptable. Not anymore. Because now we have agents, AI systems that can:
Translate intent into action
Remember our preferences
Filter noise from signal
And handle transactions safely, behind the scenes
Unlike traditional chatbots that respond to predefined inputs, or RPA bots that follow rigid workflows, agentic AI brings something fundamentally new to the table:
It reasons.
It doesn’t just act, it decides how to act, using context that evolves in real time. That’s the core shift. Execution is no longer blind, it’s adaptive. And when these agents are paired with digital wallets, trust frameworks, and protocols like MCP, something interesting happens.
Friction doesn’t just drop, it reverses.
The effort required to not use an agent becomes greater than the effort of letting the agent act. That’s the tipping point, when the technology disappears, and the behavior becomes default.
And when that shift occurs, entire markets start to behave differently.
In this new environment, travel sites, and indeed any commerce-facing platform, stop optimizing for clicks or conversions. They begin optimizing for agent compatibility. Success is no longer measured by visual design or UX tricks. It’s measured by whether an autonomous agent can find you, understand you, and transact with you, all without human intervention.
That’s the new competition - becoming a preferred node in the agent’s network. Not just visible, but trusted. Not just accessible, but actionable.
And the businesses that make this shift first won’t just benefit from being early. They’ll get something even more powerful: first access to intent. Because when a user trusts their agent, they’re no longer clicking. They’re delegating.
And by the time the human sees the option, the agent has already moved.
Thanks Gam.
Another refreshing reminder of why and how AI Agents are going to have a profound impact on the digital customer experience. Whether businesses are ready or not.
And it won’t just be travel and hospitality. We can expect this disruption across all customer touchpoints, across all sectors.
Next week, we’ll dive into Part Two, to unpack:
What actually happens when agents talk
Inside the agent-to-agent interaction
How digital wallets and data stores power personalization
Preparing for a world where agents are your customers
I can’t wait.
There are more deep dives like this coming soon, with experts from across the Customer Futures Network getting to the heart of more big topics. Like where Personal AI meets regulation, where payments meet digital wallets, and updates from the EU’s Digital ID programme.
All of it, of course, is about understanding the future of being a digital customer.
So thanks again for coming back to the Customer Futures newsletter.
Until next time,
Jamie
And that’s a wrap. Stay tuned for more Customer Futures soon, both here and over at LinkedIn.
And if you’re not yet signed up, why not subscribe: