Cognitive Storytelling Time

In working on the design of IBM Watson Work Services, it’s sometimes difficult to communicate and get to the heart of how this thing works and more importantly, how to conceptualize the design so that people can really get it.

Stories, especially ones that carry real world analogical treatment of the subject are great to allow people to see that “aha” moment for themselves.

So here’s the story for the cognitive model of Watson Work Services.

Imagine that you have a special kind of assistant working with you. Someone you have absolute trust of and, well, they are really very focused…in an almost “idiot-savant” kind of way. This guy, let’s call him John, he has a focus, paying attention to Tasks, Actions, Questions and Commitments. That’s it. You say to him, “nice weather” and it’s like he’s not there. However, you say, “I wonder if it’s going to rain” and he lights up, like someone just hit him with the de-fib paddles and shot him full of adrenalin.

Anytime, anywhere in any communication you have: one the phone, emails, chats, etc — he lights up. If he smells a question, a task, a commitment or an action, he gets real animated. Otherwise he’s just not even there.

So of course, if you step away from your communication for a while, say a weekend, and then come back to things, this dude can let you know about every question, every task and every action that came up.

How sweet would that be?

In it’s essence, this is what Watson Work Services does. Every message that passes through Watson Work Services has Watson reviewing it and this Watson knows all about Questions, Commitments, Tasks and Actions.

You can do a lot of powerful things if you can tap into this kind of understanding of conversations between people in teams. And do more. Act upon this understanding.

Find answers to questions, assign tasks or even carry them out and act…directly upon any actions requested (or make it super easy for someone to act). According to David Allen, to be super productive when requests to act come to you, you need to immediately:

  1. Decide you can ignore, or
  2. Decide you can delegate, or
  3. Decide you can defer til later, (just note when and what your next step should be), or
  4. Decide it’s something you can do in less than 5 minutes and just do it now.

You can build pretty powerful app with Watson and this basic logic flow.

Now if that was the end of the story, it would still be pretty cool, but there’s an interesting twist we can add to this stuff. Let’s say you’ve got a few other friends, like John but different. Let’s say you have a friend, Joan. She happens to be very focused on things that sales people worry about…opportunities, deals, closing deals, etc. But again, like John, anything else falls on deaf ears.

Joan can only see the world in terms of these things and she can get pretty excited about them. Say something about “moules et frite” and it’s like she doesn’t care at all.

With Watson Work Services you can take the kind of understanding of conversation Joan has and encode it with Watson Conversation, then add it to your Watson Work Services app. In doing so, your App has both John and Joan listening to the conversation and paying attention to those things they know about, what it is they focus on.

This is the model for Watson Work Services. Any message that goes into the system, if it’s a Question, Task, Commitment or Action, the general Watson classifier that’s built into Watson Work Services will add a “focus” annotation with this information, (e.g., this is a question)

Now when you create a custom Watson Conversation (like Joan), and you add that to your app, it will create a focus annotation for any message that has anything with sales (opportunities, deals, etc).

In essence, these classifiers are trained to see the world through a very focused sort of lenses, and you can turn that maniacal focus into powerful apps and bots that work in Watson Work Services.

I hope this story helped you to see how Watson Work Services employ cognitive capability to understand the conversation.