Technology Innovates. Markets and Strategy Evolve Through Standards.
Innovation creates new possibilities. Standards determine which of those possibilities become practical, interoperable and widely adopted. Computing has repeated this pattern for decades. As new ideas emerge, different vendors develop their own approaches to solving the same problem. Over time, common standards begin to appear, allowing independently developed products and services to work together. That interoperability encourages wider adoption, creates larger ecosystems and provides organisations with greater confidence when making long-term technology investments.
The internet itself is built on this principle. Standards such as TCP/IP allowed different computer networks to communicate. HTTP established a common way for web browsers and web servers to exchange information. USB gave hardware manufacturers a standard interface for connecting peripherals, allowing an enormous ecosystem of compatible devices to develop. In each case the standard defined how products could work or be used together.
Artificial intelligence is entering a similar stage of evolution. Large language models, AI assistants, software platforms and business applications are advancing rapidly, while new standards are emerging to help them communicate securely and exchange information in predictable ways. Some of these standards are already being adopted, others are still evolving, and many questions around interoperability, identity and trust are still being answered, but some trends are emerging.
For business and technology leaders, this is the strategic story. The opportunity lies less in any individual product than in the emergence of common ways for systems to collaborate. As those standards mature, organisations will gain new options for exposing services, integrating applications and interacting with customers through an expanding range of digital channels and possible services or service formats.
A Simple Mental Model
Imagine asking your AI assistant:
"Book me a round of golf next Saturday morning within a half-hour's drive, invite my usual four-ball, and avoid courses we've played during the last six months."
From your perspective, it's a single conversation. Behind the scenes, however, several independent systems may need to cooperate. Your assistant may search for suitable golf courses, compare availability, check your calendar, identify your regular playing partners, verify your membership status at participating clubs, confirm green fees, make the booking and send invitations.
Each of those tasks may involve a different organisation with its own software, business rules and data. The golf clubs maintain their own booking systems. Your calendar belongs to a different platform. Payment may be handled by another provider, while your membership details are stored elsewhere and each of your regular four-ball has their own AI agent and will need notification and to confirm. Rather than one application containing all of this information, multiple systems work together to complete a single request.
This represents an important shift in the way online services are delivered. Traditionally, users interacted directly with individual websites or mobile applications. Increasingly, software will be able to coordinate those interactions on the user's behalf, communicating with the underlying business systems and presenting the result as a single, seamless conversation.
Making that possible requires different ways for software to communicate. Some technologies allow an AI assistant to retrieve information from systems within its own organisation. Others enable software developed by different organisations to exchange information securely. Existing APIs continue to provide deterministic access to business functions such as payments and transactions. Together, these approaches form the emerging architecture behind AI-enabled online services.
Understanding the Terminology
The communication technologies behind AI-enabled online services are designed to solve different communication problems. Although they're often mentioned together, they serve different purposes and, in many cases, will work alongside one another.
Application Programming Interfaces (APIs) have been part of modern software development for many years. An API allows one software application to request information or perform a specific action in another. They are widely used for tasks such as processing payments, checking stock availability, creating orders or retrieving customer information. APIs remain one of the most important ways for business systems to communicate and are likely to continue playing that role.
Model Context Protocol (MCP) focuses on communication between an AI model and the systems or information it needs to access. Rather than searching the internet for an answer, an AI assistant can retrieve information from business applications such as an ERP system, CRM platform, document repository or booking system. This allows the assistant to answer questions or complete tasks using current, authoritative business data.
Agent2Agent (A2A) extends communication beyond a single organisation. Instead of interacting directly with internal systems, an AI agent communicates with another AI agent representing a different organisation. Returning to the golf example, a booking service could ask participating golf clubs about tee times, pricing or membership benefits, receiving structured responses from each before presenting the best options to the user.
Viewed together, these technologies form complementary layers rather than competing alternatives. APIs expose business capabilities. MCP gives AI assistants access to information and systems within an organisation. A2A enables AI agents representing different organisations to cooperate. As online services continue to evolve, organisations are likely to use all three, each where it delivers the greatest value.
Why Standards Matter
The technologies themselves are only part of the picture. Widespread adoption depends on common standards that allow independently developed systems to communicate consistently, establish trust and exchange information with confidence.
Returning to our golf example, imagine your organisation runs customer golf days throughout the year. You ask your assistant to show you your next five events. The answer may require information from several golf clubs, each using different booking software and maintaining its own records. Every club needs to understand the request, identify the correct customer, determine what information that customer is authorised to access and return the answer in a format that can be understood by the requesting system.
Those requirements extend far beyond simply exchanging messages. Organisations need common approaches to establish identity, authentication, authorisation and the description of business information. A golf booking, a hotel reservation or a purchase order may represent different business processes, but each requires participating systems to agree on the meaning of the information being exchanged and the level of trust established between them.
This is one reason the industry is investing so heavily in emerging standards. The objective, more than simply to connect AI systems, is to create an ecosystem in which independently developed software can collaborate securely and reliably across organisational boundaries. Many of these standards are still evolving, and it will take time for best practices to emerge, but that process of standardisation has accompanied every significant advance in connected computing. Payments alone must meet multiple payment type requirements and naming conventions, and users must be able to track payments and indeed errors should they happen. A common language will be needed to enable that, as well as AI agents. Today that standardisation isn't yet in place.
For technology leaders, the strategic question is therefore less about selecting a particular protocol and more about preparing their organisation to participate in an increasingly connected digital ecosystem. The standards will continue to mature; the organisations best positioned to benefit are likely to be those with well-managed data, modern business systems and an architecture designed to integrate rather than operate in isolation.
What This Means for Business
For most organisations, the emergence of AI-enabled online services is an architectural consideration rather than a technology project. One that must be carefully considered because it will need to be maintained. The underlying business capabilities remain the same; the ways in which customers, partners and software interact with those capabilities are expanding.
A customer may continue to visit your website to learn about your products and services. A mobile application may use APIs to complete a purchase in a controlled and predictable way. An AI assistant may retrieve information from your internal business systems using MCP, while another organisation's AI agent may communicate with yours to coordinate bookings, verify availability or exchange business information using A2A. Each interaction reaches the same underlying systems through a different communications interface.
That makes the quality of those underlying systems increasingly important. Well-managed data, clearly defined business processes, modern ERP and CRM platforms, secure identity management and thoughtfully designed integrations create a foundation that can support new channels as they emerge. Organisations that invest in these fundamentals are well positioned to adopt new standards without repeatedly redesigning their technology landscape.
The strategic opportunity is therefore broader than artificial intelligence itself. Businesses are moving towards an environment in which information, services and transactions are expected to be available through multiple channels, whether those channels are websites, mobile applications, partner platforms or AI agents, invisibly integrated or otherwise. The organisations that adapt most successfully are likely to be those that expose their business capabilities in secure, well-structured and interoperable ways, allowing customers and partners to interact through whichever interface best suits their needs.
The technologies discussed in this article represent another stage in the continuing evolution of online services. As standards mature and adoption increases, new forms of interaction will emerge alongside those already in use. For technology leaders, the objective is not to predict which interface will become dominant, but to build an architecture capable of supporting them all.
Conclusion
This is likely to mean that things like assistants might be much smarter and more prevalent.
Often built in to websites as opposed to independent services, but also available as personal assistant services.
Imagine asking Alexa:
"Organise a weekend away for our anniversary. Find a four-star hotel within three hours' drive, arrange dinner for Saturday evening, book a spa treatment and a game drive, then show me the best itinerary."
After reviewing the options, all that's left is:
"Perfect. Go ahead and make the bookings and add everything to our calendars."
To fulfil that request, Alexa may need to communicate with hotels, restaurants, mapping services, payment providers and your own calendar or e-mail system, coordinating information from multiple organisations before presenting a single answer. Behind that seemingly simple conversation, AI agents, APIs and other emerging standards may be working together to complete the task.
The hotels are still likely to have websites. You'll probably still want to browse photographs of the rooms, explore the facilities views and surrounds, compare locations or read reviews before making a final decision. Those are visual, information-rich experiences that websites continue to deliver exceptionally well. The planning, coordination and booking, however, may increasingly become conversations between software acting on behalf of people and organisations.
That illustrates the broader direction in which online services are evolving. Business capabilities remain remarkably stable; organisations continue to sell products, provide services, manage bookings and process payments. What changes are the interfaces through which customers and software access those capabilities. Websites, mobile applications, APIs, AI assistants and Agent2Agent services each become another way of interacting with the same underlying business systems.
For technology leaders, the strategic opportunity lies in building organisations that are ready for that evolution. Modern business platforms, well-managed data, secure identity and an architecture designed for integration provide a foundation that can support both today's interfaces and those still emerging. The standards will continue to evolve, but the objective remains the same: making it easier for customers, partners and software to interact with your business through whichever channel best serves the task at hand.
One consequence of this evolution is likely to be a gradual change in how customers discover products and services. As AI assistants become another way of finding information and completing tasks, the relationship between search engines, websites and AI agents is also likely to evolve. That topic deserves a closer look, and we'll explore what it could mean for search, discoverability and the future role of platforms such as Google in a future article.