Interoperability breakthrough: how MCP becomes the universal language of business AI

Model context protocol (MCP) November 2024. During its seven months of existence, MCP seems to have become The choice of the winning protocol for the AI ​​industry. Despite the number of companies announcing MCP servers, MCP is technically not a standard.

Companies that create MCP servers believe that the establishment of the infrastructure for interoperability should start now. Many consider MCP as one of the main protocols, if not the potential winner, for the aging ecosystem.

MCP is one of the few protocols that have emerged as a means for agents built using a language model or frame to interact with a different framework. The idea behind MCP and other protocols like Agent2Agent de Google and its partners, Agntcy de Cisco and a collective of companies, and independent research groups like Loka, is to find and establish a standard of interoperability that everyone follows.

“My point of view on the real reason why interoperability and the use of tools have really emerged in the past six months approximately because, as with the mobile phone or with anything else, I think we finally reach a critical level of capacity that LLM have to use these tools,” said Jeff Wang, co -founder of the research app on the web APE, in an interview with Venturebeat.

He added that interoperability has a moment now because the models and tools it will governing are finally powerful – and useful – enough for it to be easier to start creating ways to connect.

In recent months, companies like Openai, Glean, Mongodb, Cloudflare, Paypal, Wix and Amazon Web Services have either opened MCP servers, or created a certain integration with the protocol. And the list increases.

MCP VS API

A large part of MCP’s attractiveness comes from the rationalization of how models interact with data and tools. Before MCP, the developers pointed out models and agents to data with APIs.

However, APIs are imperfect connectors, especially for agents who access data to automatically perform tasks.

Ben Flast, product director at Mongodb, said MCP provides much more control and granularity for organizations and agents.

“It’s the really powerful thing about MCP,” said Flast. “You take MCP and put it above the context you have, so you now have a very fine method with which you can control and expose the capabilities you need.”

Unlike APIs, organizations can configure their MCP servers with personalized instructions indicating what agents may or cannot access. The server can “ask” an agent his identity and determine if he can press information on the MCP customer side. Companies have more word to say about what external agents can access their side, giving MCP more managementality of the company.

Batchu sagar, co-founder and CEO of the API Speakeasy tool company, said that MCP transforms the work interface and the API into a cat interface. He said that MCP makes him if speakeasy and that his customers do not need to rewrite or manually keep the APIs constantly.

“It is very natural for speakeasy to go build MCP support and let customers create good MCP servers. MCP servers, the better they are, the better they work,” he said.

Industry is heading for MCP

AI announcements are divided into three buckets: new models, launch of AI agents or agent libraries, and now, the management of the MCP server. Companies say it is a sign that industry could choose a winner.

Yaniv same Haim, director of technology of the WIX website manufacturer, told Venturebeat in an email that MCP aligns with the objectives of the company because he thinks that MCP can act as a “bridge” for its IA development workflows.

“Wix has chosen the MCP model in particular because it aligns with the change in industry to development fueled by LLM, where intelligent interfaces rich in context are essential,” said Haim.

He said that the WIX MCP server “allows users to interact with Wix via tools like Claude, Cursor and Windisurf, directly from Ides or cat interfaces.”

In case of doubt that MCP, and to a certain extent A2A, is adopted by industry, these can be erased by the support of large companies of the protocols. Microsoft Satya Nadella’s CEO, for example, has approved both in a post X saying: “Open protocols like A2A and MCP are essential to allow the agentic web.” The CEO of Google, Sundar Pichai, also gave the enthusiastic approval of the MCP.

Naturally, some companies show caution and wait a little longer before building an MCP server. CTO Shawn Malhotra rocket companies told Venturebeat that they saw the potential of interoperability standards and build infrastructure to support them. However, Malhotra plans to wait a more critical mass before kissing MCP fully or other protocols.

“We believe that this notion in a standard way of exposing tools to agents is a really powerful paradigm. So things like MCP will allow us to enjoy even more advantages in the future,” said Malhotra. “We will experience internally and have an exhibition of tools based on MCP-Server that I think it is only a matter of time before these things happen in production.”

There could be several standards at first

Companies such as exams, confluence, merge and others are all on MCP, but they also understand that some customers can use other protocols for agents’ communication.

For many companies, MCP will be one of the many protocols they take care of, as their customers decide what agent interoperability and communication methods to be used.

Other companies said they found value in the execution protocols targeting different levels of stack of agent technologies. Walter Sun, SAP Global Head of Business Ai, said: “There are obviously many opportunities and many levels of communication, and I think that high -level agent’s liaison communication.”

Without a doubt, the agentic ecosystem will have standards of interoperability, which could arrive soon. The growing adoption of MCP, for the various reasons, of many companies, proves that the demand for standards is only growing.