AI Strategy

Why AI's Biggest Players Just Agreed on Open Standards

· 6 min read

If you missed the headline last week, here's the short version: Microsoft, Google, OpenAI, and Anthropic — four companies that spend most of their time trying to eat each other's lunch — just agreed to work together on something. They've formed the Agentic AI Foundation, backed by the Linux Foundation, with the goal of setting open standards for how AI agents connect to apps, data, and services.

Your first instinct might be to file this under "interesting tech news, not my problem." But this one's worth a few minutes of your attention, because the practical effect on business software could be significant — and it's starting to happen faster than most people realise.

What Actually Happened (Without the Press Release Fog)

The centrepiece of this announcement is a protocol called MCP — Model Context Protocol. It was originally built by Anthropic (the company behind Claude) to solve a specific problem: how do you give an AI agent a standardised way to connect to external tools, databases, and services? Think of it like a universal plug format for AI.

Anthropic has now donated MCP to the new foundation, making it vendor-neutral. That means no single company owns it. Google, Microsoft, OpenAI, AWS, and others have all signed on to support it. The Linux Foundation — which has a long track record of stewarding open-source infrastructure (Linux itself, Kubernetes, etc.) — is managing the whole thing.

If you want more background on MCP specifically, we've covered it before in this breakdown of MCP as an open standard.

The USB Analogy (Because It's Actually Perfect)

Before USB existed, every peripheral — printers, cameras, keyboards — used a different connector. Manufacturers made proprietary ports, and you needed a different cable for everything. Then USB came along and everyone agreed on one standard. Suddenly, any device worked with any computer.

That's almost exactly what's happening here with AI agents.

Right now, every AI tool connects to external apps in its own way. If you want your AI assistant to read your CRM, pull from your inbox, update a spreadsheet, and check your calendar, you typically need four separate integrations — each built and maintained differently depending on which AI platform you're using. When you switch tools, you often have to rebuild everything from scratch.

MCP becoming an open standard is the industry saying: let's all use the same connector. Build an integration once, and it works across any AI agent that supports the standard. Less rebuilding, less lock-in, more things just working.

Why It Matters If You're Running a Business

The most immediate practical benefit is durability. If you've built any AI automations in the last year — even simple ones like summarising emails or pulling data into reports — you've probably had the experience of something breaking when a tool updates its API, or when you switch from one AI assistant to another.

Open standards reduce that fragility. When your CRM, your project management tool, or your accounting software builds an MCP integration, it works with any compliant AI agent. You're not betting on one vendor's ecosystem staying intact.

The second benefit is breadth. Right now, the number of apps with native AI integrations is limited. As MCP adoption spreads — and with Google, Microsoft, and OpenAI all behind it, adoption will spread — more of your existing software will start connecting to AI natively. You won't need custom-built middleware for every connection.

What This Means If You're Already Using AI Tools

If you're using AI tools today — even in a basic way — here's the practical read:

For a broader look at how vendor lock-in and interoperability have been shaping AI tool decisions, this post on AI tool interoperability and vendor lock-in is worth reading alongside this one.

The Bigger Picture: Why Competitors Would Do This

It's worth asking the obvious question: why would companies that compete this fiercely agree to standardise anything?

The answer is that all of them benefit from growing the overall market. Right now, a huge chunk of the potential value from AI agents is locked up because connecting AI to real business systems is still too complicated. If open standards lower the barrier for businesses to adopt agentic AI workflows, everyone wins — more customers using AI, more revenue across the board.

It's the same logic that led competing tech companies to support USB, Wi-Fi, and Bluetooth. The standard creates the market. Then they compete inside it.

The formation of the Agentic AI Foundation isn't a kumbaya moment — it's a calculated bet that a bigger pie is better than a larger slice of a smaller one.

For business owners, the signal is clear: agentic AI is no longer a future concept. When the four largest AI companies align on infrastructure standards, it means they all expect agent-based workflows to become mainstream business tooling — and soon. We've been writing about this shift for a while; the move toward AI agents as everyday business tools is accelerating, not slowing down.

What to Actually Do With This Information

You don't need to take any action today based on this announcement. MCP adoption will roll out gradually, and you'll see it appear as updates inside the tools you already use rather than as a big switch you need to flip.

But there are two useful takeaways for how you think about AI in your business:

  1. Don't let "what if the standard changes" stop you from building. The risk of fragmentation just went down significantly. The tools you build on now are more likely to be durable.
  2. Start thinking in terms of connected workflows, not isolated tools. The direction the industry is moving — AI agents that connect natively to your CRM, inbox, calendar, documents — rewards businesses that have started thinking about how their tools fit together. The infrastructure is being built to make this easier. Get ahead of it.

The boring infrastructure announcements are often the most important ones. USB wasn't exciting when it launched either.


Sources

This article is grounded in the following reporting and primary-source announcements.

Continue Reading

Related articles worth reading next

These are the closest practical follow-ons if you want to go deeper on this topic.

Need help deciding what to build or teach first?

We help teams choose the right next step, whether that is training, workflow design, or a system built for a specific business problem.

Book a call See services

This article was reviewed, edited, and approved by Tahae Mahaki. AI tools supported research and drafting, but the final recommendations, examples, and wording were refined through human review.