A Great Simian or just a Monkey

watson dialog

cognitive collaboration

Cognitive Business Part 1 – Internal Collaboration

I am starting with an old passion of mine, Internal Collaboration, knowledge and information sharing as well as finding the right expertise within the organisation. I will try to be concrete and skip the marketing buzzwords. This is the first in my series of examples of real-life Cognitive Business.

I started this series since sometimes this cognitive business thing is hard to grasp without real-life examples that all of us personally can relate to. So, lets jump right in.

Imagine having a question about one of your company products or any question that is an internal only question. Today you might search the intranet, often without result, you then take to your nifty internal collaboration tool, might be Slack, Yammer or whatever tool you have. If you are lucky and after a few interactions you might get an answer to your question or at least a part of the question.

Cognitive Internal Collaboration

Now lets imagine a world where there is a new friend on your list in the tool called Filippa. Filippa is a clever girl, she actually remembers everything and she keeps track on all new things that are created or stored internally and to some extent also externally. When you put your question on the internal network you can also adress Filippa. She will then go through all of the information she has and present you with what she think is the best 5 answers, who has created the info and also some evidence to why she thinks those five pieces of info is the most important. If a complex answer she will bundle it and present it in a nice way for you, still with evidence and who is the expert in the area and who has created the info.

Immediate Benefits

This will not only provide some really good answers to your question, this will also contribute to the conversation that might continue afterward. This since many other might chip-in on the conversation you have with Filippa and bring even more value to the answer. Filippa, as the caring and non-selfish individual she is, will also give credit to the original authors of the info and naturally @mention them in her reply, so now the conversation can really elevate and the end-result you will walk away with will be something like this:

  1. The answer will be of very high-quality
  2. Given to you within seconds
  3. Expertise on the topic identified and also invited to the conversation to further elaborate as well as giving credit to the expertise.
  4. Not only did you learn, so did Filippa. She added this conversation to her “knowledge” and is now even smarter and ready for even trickier questions within the area of expertise.

Just imagine the amount of time you spend searching for the right information or expertise within your organisation, that is a huge pain and cost for todays organisations.

In this first part of the “Cognitive Business Examples”-series we have touch what many consider the holy grail of collaboration. I hope this use-case / example fills some gaps and that the potential is obvious?

In the next part of this series we will look at how Marketing could work in a cognitive business.

This is the first post in my Real-life Cognitive Business series. Below is part 2 and the other parts will be added as they are posted.

Cognitive Business Part 2: Marketing

The photo is from Harstena in Gryts Archepelago in the Swedish east coast. A true favorite place on earth. The pump could represent the pumping and flowing of information that is described in the post.

watson assistant

Watson as an Assistant

When talking IBM Watson the interaction between the user and Watson is mostly done via a “chatty”, Watson assistant like, dialog interface and in natural language. This has created a popular use-case for Watson, the personal assistant.

A dialog is nothing new, there are tons of assistant apps and products out there, it is a big trend, so why is Watson so unique?

The uniqueness is mainly in the department of learning and understanding. The dialog in mainly managed by a service called Dialog. Dialog is actually pretty dumb, or honestly, plain stupid, just like any programmed dialog, nothing new there. What is new is when you add other Watson services to the Dialog experience and enable Watson to learn from the interaction with the human in the other end.

Watson learns from every interaction

If we take a quick look at the list of examples at the end of the post, we quickly see that it is a wide array of verticals that use Watson as an assistant. The greatness surfaces when you realize that Watson can learn domain specific language. An Oncology doctor and a person that want to buy a jacket from North Face do definitely use different and domain specific language. Watson will, overtime, learn this language. Initially Watson will be pretty rigid in terms of what kind of dialog that can be had, it is mainly following the script in the Dialog XML that is used. With some training (look at the Natural Language Classifier as an example) this will improve rapidly and Watson will learn your domain without you lifting a finger.

That is not all that Watson understands how you communicate, he also helps build a better knowledge within the domain, every interaction with Watson will also make Watson smarter. Therefor the interaction is an important part of Watson. It is not necessary though and Watson can learn in may differnt ways. When training Watson, large chunks of data can be uploaded (and updated continously) to Watsons corpus. Watson can read 800M pages per second, that is a bit more than me, but I am a slow reader. When the data is entered into the corpus, Watson need to be trained, now we are back at the “chatty” interface again, but this time only to tell Watson if things he says are good or bad answers The training can be done in several ways.

So again we are back to the fact that Watson actually learns and communicates a lot like a human. Cognitive computing that is.

More use-cases with Watson as an assistant?

To start, I am not a believer in Watson replacing customer service or being the only interface for customers. Watson is very capable of managing specific processes or specific questions like “I have lost my password” or “When do I have to submit my yearly tax report?”and based on those questions provide information and actions. It can definitely replace quite a few of these shitty “Tell me what you want and I will direct you to a customer support agent”-solutions. The great thing there is that Watson can act in all channels with the same solution. Social media, call-center, web, in-store etc. It would cut cost radically (I know quite a few companies are looking at this incl customer support companies).

The greatness does in my oponion not lay in the interaction, it lays in the answers that Watson can provide. A customer service reperesentative, sales person, communication specialist etc do not know every document that might provide an answer, Watson actually knows literally every document. Since Watson also can answer questions and provide evidence for the suggestions he replies with, Watson is a tremendouse help for any individual in an organization that needs to answer a question based on unstrucutred data. Therefor Watson can be a multi-layered implementation for any assistant solution build on Watson. It can help in the first interaction with different tasks as well as direct the client to the right section, person or information. Primarily I would say Watson will provide information to representatives that engage with clients, this to provide the best answer possible and then provide feedback back into Watson. This was a bit of rambling, but I will leave it as is.

Watson Assistant – examples

A few examples of these use-cases are:

IBM already have created best practices within Telco and Insurance. Both are based on the product Watson Engagement Advisor, which is an out-of-the box Watson Assistant. It seems IBM want to start every Watson case with a Cognitive Value Assesment. Have not managed to find that much info on these CVA, except that they should contain these bullets:

  1. Define use case and user personas
  2. Develop benefit model
  3. Map out cognitive journey
  4. Generate stakeholder buy in and confirm commitment

After the CVA a go / no-go decision is made.

What are your thoughts on the Watson Assistant use-case?

Photo by Christian Bucad

Powered by WordPress & Theme by Anders Norén