The emergence of knowledge automation
This is the fourth post in our guest blog series by Esteban Kolsky of ThinkJar. See the original post for an introduction and a list of other posts in the series.
Starting in the early 2010's, the demand for new models for KM was almost deafening. Three trends made the need for new KM felt in organisations:
- The revolution in customer interactions led by online communities and social media
- The sheer amount of knowledge known to exist in communities – and the subsequent rise in Google-initiated searches for answers
- The discovery that existing KM was not only very expensive to maintain, but also not very effective.
The search for new models began to take a toll on (then) existing vendors who went in different directions. Some of them became search and traditional KM vendors and perpetuated the model of knowledge-in-storage, but a few brave souls began to play with AI technologies and concepts and to look for answers elsewhere.
From these few vendors and practitioners, we saw the emergence of knowledge automation.
We came across knowledge automation when we exhausted the ability of our then state of the art KM system to find information quickly. We got better speed, but also amazing accuracy in finding the right answer.
The concept of knowledge automation is an evolution from the old models of store-and-retrieve knowledge. While there is value to that model, in a fast-moving, more contextual-focused world the model was not delivering. Latency was increasing, differentiation between entries was not achieved, and taxonomies were rendered useless when incorporating the myriad variables that organisations were collecting as part of their Big Data and Customer Engagement initiatives.
A query that was created to use a few, in some cases one, simple keyword inputs is not suited for a complex situation where the organization is trying to find the one answer to a problem. Making acute use of the key characteristic of computers and AI, repetitive processing of information to augment its value, a knowledge automation project focuses on pre-fetching the answer as more and more information becomes available. It performs constant, iterative searches on the knowledgebase with increasing understanding until the right answer is found.
Take the example of a person approaching retirement age and looking for information on how to balance their pension portfolio. The knowledge solution would start with the content of the question: how to balance a portfolio. Then it would refine its understanding of the question to focus on retirement funding. Then it would further refine its understanding using data from the customer’s account, enabling it to find the one answer that best fits both the content and the context of the question.
This continuous refinement of the search context is what knowledge automation is all about: automated pre-fetching of information that is continuously optimised and personalised as more data is made available, zeroing in on the answer that matters.
The value of this method, still in its infancy, is the ability of the AI-augmented solution to focus its understanding by processing millions of potential variables, the ability of the KM system to find further refined knowledge in record time, and the ability of the knowledge automation solution to put both together. And while issues remain on adoption at scale, the few practitioners that I spoke with were absolutely convinced that this is the future of KM and AI working together.