Microsoft is making good progress on the inter-compnay collaboration journey with their AI agents. As organizations need to collaborate more and more effectively within intricate ecosystems, Microsoft is working on AI-powered solutions to dissolve the longstanding constraints and enable a new era of partnership.
At its heart the project involves creating AI agents that can work outside of organisational silos. These smart systems are built to be able to talk to, exchange data with and perform tasks alongside AI agents created by other firms.
Microsoft is enabling this interconnectivity through its use of open standards such as the Model Context Protocol (MCP), an initiative also sponsored by Anthropic. MCP serves as an universal language that enables the AI agents, born in different sections of the multi-verse, to comprehend and interact with one another.
“Picture an ‘agentic web’ of A.I.-powered services that can work with one another — in the same way that human agents collaborate and work with one another across disciplinary boundaries,” Kevin Scott, Microsoft’s chief technology officer said in a recent tech briefing. “This is the future we are building, where companies can use AI assistants that specialize in specific tasks and can work with brands across different ecosystems, giving them more freedom and flexibility.”
One of the challenges in recent AI interactions is the problem of no memory and no context kept. To combat this, Microsoft has come out with a technique it calls “structured retrieval augmentation.”
This enables AI agents to summarize and remember important information from previous interactions and accumulate context sensitivity.
So if an AI agent from one company interfaces with a agent from another company it can remember past conversations, making the interaction more useful and relevant and reducing the need to repeat the same conversation, it also increases the context across an conversation between agents.
Such flawless inter-company AI cooperation could pay huge dividends. With AI agents orchestrating logistics and inventory activities in real time across many partners, supply chain management can be dramatically improved.
Two-party projects could use AI to simplify project management, disseminate data insights, and automate intricate workflows. As for research and development, it is imagined that AI agents, acting as digital liaison officers or ‘diplomats’, would make sure that findings are shared and innovation enhanced across cooperating institutions.
But there’s no easy road to widespread use of inter-company AI agent cooperation. It is critical to secure and protect shared data between organizations. The need for strong security protocols and governance frameworks is therefore paramount to establish confidence and compliance.
Combining AI agents with the already deployed legacy systems in various companies is also a major technical challenge. In addition, communicating clear communication protocols and standards associated with these AI interactions will be necessary for effective joint efforts.
Despite all these hurdles, Microsoft’s willingness to create interoperable, context-aware AI agents introduces a tectonic shift in how companies will work together. With the development of “agentic web,” Microsoft hopes to enable organizations to collaborate more effectively together, move the needle on innovation, and harness a new level of productivity in an increasingly networked world.
The features announced in advance of Microsoft’s Build 2025 conference suggest clear intentions for a future when AI is not a solitary teammate, not only within a single company, but an entire industry.