IT之家 On October 15, tech media NeoWin today (October 15) published a blog post, reporting that in enterprise internal knowledge-intensive scenarios, Microsoft has achieved a technology breakthrough for AI agents such as Copilot: without upgrading the underlying large model,the AI agents' response speed has been increased to 4 times the original, while token consumption has been reduced by about 30%.
According to the blog post cited by IT之家, Microsoft stated that in enterprise office scenarios, project information is often scattered across emails, Teams meeting records, and technical documents. In the past, AI agents had to retrieve and process these fragmented source data one by one, resulting in high latency and high computing costs.
The latest technology is built on SharePoint and equipped with Microsoft's brand-new Work IQ knowledge compilation engine, which can automatically transform scattered information across organizations into a structured knowledge base that AI agents can access directly and efficiently.
In terms of applicable scenarios, it mainly covers cross-document question answering, team context retrieval, organizing meeting key points, summarizing project status, retrieving internal processes, preparing customer communication materials, or generating first drafts based on existing materials.
The feature is now rolling out to Frontier Preview tenants. Organizations can add the knowledge base to existing SharePoint sites, including sites that support Teams channels and Microsoft 365 groups.
This preview feature is currently only available to users who have enabled pay-as-you-go billing, and administrators can manage access and spending through the Microsoft 365 admin center. Microsoft also plans to extend it to Copilot Studio custom agents and develop automated maintenance capabilities so that the knowledge base stays updated with enterprise content.