Everybody wants effective knowledge management in their enterprise. Nobody wants to exert the effort or commit the resources necessary to make KM happen. Creating and capturing knowledge is a pain and the effort is rarely recognized or rewarded. Once expertise is pulled out of someone’s head and deposited in a knowledgebase, it is usually out of date before anyone consumes it. Besides, who has the time to curate content when there are real deliverables to attend to? Knowledge management is a nice idea, but for most organizations and individuals it seems to be more trouble than it’s worth. Microsoft might finally be in a position to change that narrative with Project Cortex.
Project Cortex pulls together multiple technologies from Microsoft research and turns them loose on the Office 365 ecosystem. As content is created and consumed, various AI-driven mechanisms process that content to extract entities, metadata and signals. These cues are compared and clustered until the system can classify it, tag it and decompose it into “knowledge entities,” a new class of object in the Microsoft Graph. Knowledge entities are grouped into AI generated topics and made available on-demand. Microsoft is attempting to realize the dream of an automatically generated knowledgebase. They aren’t there yet.
The topic pages generated by the AI undergirding Project Cortex are better thought of as topic stubs or article frameworks. They pull together, organize and present information culled from content across the environment, but the resulting pages still need a human touch. To be truly useful, subject matter experts will still need to flesh out the topic pages with context and color that isn’t automatically extracted from the native content. This isn’t a criticism. It is an example of the centaur principle. AI works better when there is a human in the loop, and vice versa. Protect Cortex can jumpstart knowledge capture, but it still takes a human curator to fill in the gaps and make it pretty.
The benefit of that jumpstart should not be underestimated. When an expert is asked to capture the tacit information, the know-how, rolling around in their heads, getting started is always the most difficult part. An automatically generated topic page, however sparse or mechanical, can provide the skeleton upon which the expert can arrange the meat of the matter. With AI pulling context and signals from content and behavior across Office 365, the topics can act as a well-informed interviewer asking probing questions and prompting further exploration. It can get the KM-curious over the hump of getting started with knowledge capture.
Perhaps even more important than the automatic creation of knowledge entities is the automatic organization of those entities into “knowledge networks.” Relationships among AI-generated knowledge entities and other entities in the Microsoft Graph are used to create linkages among topics, content and most importantly people. This is the heart of knowledge management. A bit of knowledge may be useful in and of itself, but when placed in the context of the broader knowledgescape of the enterprise its utility and value is substantially increased. When that knowledge is associated explicitly with the people who are stakeholders in that topic, it can become transformative.
There is a lot more to Project Cortex, some of it already in private preview, much of it still “on the roadmap.” I will cover many of these capabilities and their ramifications in subsequent posts. If Cortex materializes as promoted, it has the potential to revitalize knowledge management, at least within the context of the Microsoft ecosystem. It could also cement the place of Office 365 as the dominant collaboration and productivity platform for the foreseeable future.