Microsoft Chief Technology Officer Kevin Scott wants
businesses to prioritize openness and simplicity, and believes the future will be built on artificial intelligence (AI) and the “agentic web” -- where the technology does more
than answer questions and collaborates with other platforms, accessing tools and conducting sophisticated tasks autonomously.
Ahead of Microsoft's annual software developer conference,
which begins today, Scott warned that a closed yet integrated approach could stifle innovation.
Similar to the concept of the internet of things, where all devices and mechanical appliances connected to the web and transmit data, Scott said AI agents can only reach their full potential if they can “talk to everything in the world.”
The company is reportedly working on AI agents that focus on memory functions.
advertisement
advertisement
Many systems today operate based on transactional elements such as shopping apps without retailing prior content, which weakens the possibility of successful future ad targeting with browser cookies.
Improving the memory of AI takes computational power that raises server costs.
This type of technology for advertisers would mean using a technique called structured retrieval augmentation, which Microsoft is exploring.
This approach enables an AI agent to capture concise summaries of conversational topic turns that effectively create a path back to past interactions.
With these advancements, Microsoft aims to deliver AI systems that can provide more personalized and context-aware assistance.
This could change how consumers interact with online content, advertising, and tools. It also could improve ad serving and media buying, as it memorizes past interactions with content and eliminate the need to track individual user behavior across websites, addressing privacy concerns associated with third-party cookies.
It could potentially replace or augment cookie-based ad targeting by using contextual information retrieval rather than relying on individual user tracking.
Google is holding its own I/O developer conference this week, and plans to highlight its open standard, Agent-to-Agent protocol designed to enable AI agents to communicate and collaborate directly across systems.
Microsoft leans toward using Model Context Protocol (MCP), which Amazon- and Google-backed Anthropic have developed and open-sourced.
MCP defines how AI models access and interact with external tools and data sources.