Skip to main content

The Invisible Revolution: Are AI Agents About to Order Our Fish and Chips (and Run Our Lives)?

By

The Invisible Revolution: Are AI Agents About to Order Our Fish and Chips (and Run Our Lives)

Imagine this: You're sightseeing in London. Your smart glasses catch a glimpse of Big Ben, and then you speak: "Find the nearest fish and chips shop, order two lots of fish and chips with a Savloy sausage, and have it delivered to my hotel room, number 200."

Science fiction, right? Perhaps not for long. This seemingly simple command encapsulates the ultimate goal of "agentic AI"—and it's a future that's rapidly approaching, igniting an intense battle among the world's biggest tech companies.

I recently had a fascinating conversation where we explored just how close we are to this reality, the massive technical hurdles, and the strategic plays shaping this invisible revolution.


From Sci-Fi Wish to Near-Term Reality


My conversational partner's scenario wasn't just a fantasy; it was a perfect litmus test for agentic AI. Here's a quick breakdown of what makes it so complex:

  • Location & Search: Easily doable today with GPS and advanced mapping.

  • Menu & Ordering: This is where it gets tricky. An agent needs to understand menus (not just keyword match), handle out-of-stock items, and interface with vastly different ordering systems (from modern APIs to basic websites).

  • Logistics & Delivery: The real kicker. How does an AI agent instruct a human delivery driver to navigate a hotel lobby and deliver to a specific room number, interacting with a concierge along the way?

While individual components exist, getting them to work together flawlessly, autonomously, and reliably in the real world is the ultimate challenge. We're talking about a multi-step, multi-domain autonomous agent—a "master agent" that orchestrates multiple sub-agents to achieve a complex goal.

The consensus? We're surprisingly close. Think 2-5 years for reliable, commercially available systems, with the "messy real-world" scenarios being the final, hardest frontier.


The "New Standard": AP2 and the Agent Economy


The biggest takeaway from our discussion was the realization that this future won't just magically happen. It requires a fundamental shift in how businesses interact digitally. A static website with a PDF menu is "invisible" to an AI agent.

The solution isn't a single, open internet standard (like HTTP). Instead, we're seeing the emergence of competing "orchestration platforms" and, crucially, new protocols.

Enter the Agent Payments Protocol (AP2). This isn't theoretical; it was formally announced in September 2025 by Google, alongside a consortium of over 60 partners including Mastercard, Visa, PayPal, and Stripe.

AP2 is the new "standard" that allows agents to securely pay for things on your behalf. It doesn't just send a payment; it creates a "Mandate"—a cryptographically signed, tamper-proof digital contract. This mandate contains all the details: who authorized the payment, what was ordered, how much, and the exact items. This brings unprecedented security and accountability to agent-driven transactions.

Imagine your fish and chip order: your smart glasses would generate a "Cart Mandate," signed by your secure device (think Face ID or a passkey). This "mandate" then travels through the system, verifiable by the chip shop, the delivery service, and your bank. No more rogue agents ordering 100 sausages!

AP2 sits atop other crucial protocols like the Model Context Protocol (MCP) (for agents to discover tools) and the Agent2Agent (A2A) Protocol (for secure agent-to-agent communication).


The Battle for Your AI Future: Walled Gardens vs. Open Ecosystems


This isn't just about technical standards; it's about control. The tech giants are locked in a high-stakes battle to become the dominant "platform" for this new agent economy.

1. The "Walled Garden" Approach (Apple & Google):

  • Apple: Predictably, Apple is building a premium, private, and deeply integrated "Walled Garden." Their "Apple Intelligence" agent, living on your iPhone and Vision Pro, will primarily interact with Apple's own services (Maps, Pay) and carefully vetted partners. Privacy is their paramount selling point.

  • Google: Google's strategy is to be the "Universal Orchestrator" for your life. Their Project Astra agent (the evolution of Gemini Live) leverages Google's unparalleled real-world data: Maps, Search, Android, Wallet, Gmail. Their "unfair advantage" is owning the entire stack needed for tasks like ordering fish and chips. AP2 is a core part of this, aiming to be the universal payment plumbing.

2. The "Horizontal Platform" Approach (Microsoft/OpenAI):

  • Microsoft: They are winning the "Agent for Work" race with Copilot deeply integrated into Microsoft 365. By controlling your professional life, they aim to expand outward, making their agent indispensable for travel planning and personal productivity.

  • OpenAI: They are tackling the problem from the "outside-in," aiming to be the "Android" or "Windows" for AI agents. Their GPT Agents and "Actions" framework allow any business to create a "skill" for their agent. Crucially, they're developing "Operator" agents that can visually read and interact with any website, even if it doesn't have a dedicated API—a powerful direct counter to Google's integrated approach.


Who's Winning?


It's a two-horse race between Google and Microsoft/OpenAI, but they're running slightly different races:

  • Microsoft/OpenAI has a strong lead in the "Agent for Work" (Enterprise) space.

  • Google is neck-and-neck in the "Agent for Life" (Consumer) space, leveraging its deep integration with real-world services.

The crucial point is that they're both building ecosystems. Google wants to own the entire journey (Maps, Search, Pay). Microsoft/OpenAI wants to own the agent platform that connects to everyone else's services, making them a central hub.


The Invisible Revolution is Here


The "smart glasses ordering fish and chips" scenario is no longer a distant dream. The foundational protocols (like AP2) are being laid, the major players are deploying their strategies, and the developer frameworks are emerging.

We are entering an era where AI agents won't just chat with us; they will act for us. They will manage our schedules, book our travel, and yes, potentially even ensure that two lots of fish and chips with a Savoy sausage find their way to our hotel room, number 200, without us lifting a finger.

The debate isn't about if this will happen, but how—and who will control the invisible threads that weave our agent-powered future together.

Labels:

Comments

Post a Comment

Popular posts from this blog

The Brain in the Server Rack: Why Biological Computers Are the Next Big Thing (And Why They Aren't Here Yet)

By
Imagine a supercomputer that rivals the world’s fastest systems but runs on the energy of a dim lightbulb. It sounds like science fiction, but in labs from Australia to Switzerland, it is quickly becoming science fact. We are entering the era of Biological Computing—using living human neurons instead of silicon chips to process information. It’s a technology that promises to solve the massive energy crisis facing our data centers, but it comes with a strange new set of problems: these computers need to be fed, they produce waste, and—most hauntingly—they might one day have feelings. Here is a look at where this technology stands today, and why you won’t be buying a "brain-powered" laptop anytime soon. The Problem: Silicon is Hungry To understand why scientists are growing "brains in dishes," you have to look at the power bill. The Silicon Reality: A cutting-edge supercomputer like Frontier consumes roughly 21 megawatts of power. The Biological Re...
Post a Comment
Read more

Quantum Leap: The Cloud's Next Frontier & AI's Ultimate Upgrade

By
The whisper of quantum computing has been growing louder, evolving from a scientific curiosity to a tangible, albeit still nascent, technology. As we peer into the near future, two colossal sectors — cloud services and Artificial Intelligence — stand poised to be both beneficiaries and battlegrounds for this revolutionary computing paradigm. The integration of quantum power isn't just an upgrade; it's a fundamental shift, presenting opportunities for unprecedented innovation alongside significant, even existential, threats. The Opportunity: A New Era of Computational Power Imagine a world where the most intractable problems of today become solvable. That's the promise quantum computing brings to the cloud and AI. 1. Quantum Computing as a Service (QCaaS): Democratizing the Impossible Just as cloud computing made supercomputers accessible to startups, QCaaS is democratizing quantum power. Companies like IBM, Google, and Amazon are leading the charge, offering re...
Post a Comment
Read more

MVNO PaaS solutions

By
How a Platform as a Service (PaaS) can enable an MVNO solution. The Power of PaaS for MVNOs In the telecommunications industry, Mobile Virtual Network Operators (MVNOs) have revolutionized the way mobile services are delivered. By operating on the infrastructure of Mobile Network Operators (MNOs), MVNOs can focus on customer service, brand building, and creating unique service offerings without the massive capital expenditure of building and maintaining a physical network. A key enabler for this agile business model is the adoption of a cloud-based Platform as a Service (PaaS), particularly one that integrates Business Support System (BSS) and Operations Support System (OSS) services. BSS, OSS, and the Cloudsim Solution A robust MVNO requires seamless coordination between its BSS and OSS.  * BSS (customer-facing) manages the business side, handling customer relationship management (CRM), billing, and service activation.  * OSS (network-facing) ensures the technica...
Post a Comment
Read more