Even common knowledge worker tasks like agreeing on meeting times are painfully difficult to accomplish with standard software. There are specialized commercial solutions like TimeBridge that offer new work flows for particular problems, but the SRI/CMU Radar research project is the first attempt at inventing a whole new approach to interacting with your information.
It uses machine-learning to bridge the gap between the unstructured information that flows through your inbox, and the rigid data needed by CRM and other systems you interact with. For example, their Virtual Information Officer acts like an expanded version of I Want Sandy's virtual assistant; you email it "Change John Doe's phone number to 555 313 7172" and it sends back an updated contact profile for you to approve. This natural language interface is designed to work across all kinds of databases, learning as it goes from which updates are rejected or accepted.
Another interesting module is the Briefing Assistant, which takes collections of emails and produces a draft summary report from them. Initially using a set of generic rules, it observes how each user edits the draft and uses that information to produce a better first draft in the future. There's a whole series of other agents, all improving the way we work with mail information.
What makes the Radar project stand out is its sheer ambition. I was lucky enough to meet up two of the team at Defrag, Dr Michael Freed and Teri Elniski. They gave me the rundown on the rigorous user testing they use to ensure that the redesigned workflows and tools allow people to do their jobs faster. One of the scenarios is actually planning a conference, trying to take over at the last minute from an organizer and sort out all the travel, room, equipment, speaker and expense planning. Having frequent and rigorous testing of all their changes has obviously let them experiment with some pretty radical approaches, without the fear that they're heading wildly off the right track.
I haven't been able to play with the tools myself, but the screenshots and documentation are tantalising. It's inspiring to see such a bold effort to reinvent the whole way we interact with our computers, I'm looking forward to seeing this technology develop into something that's on all our desktops.
Pete Warden's blog 