Skip to main content

Augmented knowledge as a discipline

This is the third 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.

Knowledge seldom is the entire answer. There are three components to generating the right information, to get to the right answer: knowledge, content, and data. 

Content is static, is “information” that seldom if ever changes (and when it does, is updated by the people who know what’s in it). Knowledge is the application of that content to a specific situation, bringing to bear the context and the intent of the interaction. Data, the third component of information, is the operational parameters used to frame an interaction. 

The answer for every question comes down to a combination of the three (see chart below) and, as you can imagine, changes every time given context, intent, time-sensitive needs and demands, and combinations of variables and operational parameters that make every interaction personalized and unique.

Equiniti presentation

Traditional KM used in enterprise is lagging: the focus is in creating content, storing it, indexing it, and retrieving it for later use as necessary. Alas, the components that personalize that content, and that truly create knowledge – content in action - are lost in the store-and-find model of knowledge-in-storage.

Knowledge management augmented by artificial intelligence implementations can generate far better results: unique simple answers to unique personalized problems. This is the sought-after goal for deploying knowledge management: finding THE ONE answer. AI can help you get closer – and, eventually, there.

There are three stages where AI can augment and deliver value.


Being able to identify the right information as to where it should be stored, how it should be accessed, how to classify it, and how to use it is traditionally done by a team (in some cases even a single person) of knowledge specialists that tag and index each article or piece of information added to a knowledgebase. There are three problems in this model:

  1. Manual, tedious process: leveraging the tools of AI we can more accurately, and automatically, tag and classify the information in relation to its usage and relative value to each solution.
  2. Slow to react to changes: it’s hard to know when to make changes to the content, the tagging, or the indexing.  Information that was useful for a specific use case today may not be the same for tomorrow.  AI can pinpoint the necessary changes, making it faster to change the content.
  3. Improper maintenance: While few users provide feedback on content that’s outdated or not useful, AI tools can quickly and efficiently show information that may need maintenance thus optimizing the maintenance processes.

AI can bring value to these problems by focusing on a better way to find the content, not just having to index it, in myriad different locations.


Once the information to be used in knowledgebases is identified, indexed and classified – either by computers or by humans – it is stored for future retrieval and use.  Although the process sounds simple enough - put an article into a knowledgebase - there is more to it and this is where problems arise.

All organizations have multiple repositories of content, data, and knowledge. Finding which of these repositories are useful to whom, whether the information must be used in different languages or in different regions, which are the most effective and efficient, which has the most flexible models, and other questions that relate to the efficiency of the storage of the information (not to mention that data is usually stored in database with very different and complex security and rights models) is not simple.

Imagine an organization that can reduce the size of their knowledgebase and all other information repositories, access the information faster, and virtually not worry about what information is valuable to store or to use quickly and discard. AI can help with that.


When a stakeholder asks a question, brings forth an issue, or needs a resolution, they will choose a channel and send a message to the organization. The organization will then need to find the right answer, utilizing all accessible information, and compose the appropriate response. 

AI can solve the first part of the equation. It understands given the right parameters and with access to the timely variables. Knowledge management, by contrast, does best the second part of the interaction – finding the appropriate response. The value AI brings is a better formed question for KM to frame a better answer (THE ONE). 

AI does not just provide a better-formed question for faster, appropriate retrieval. It can also function as an early indicator for potential problems in the retrieval process by noticing and monitoring the follow-on activities after an answer is given. 


The most contentious part of KM as implemented today is maintenance. 

Maintenance does not get the budget, focus, or resource allocations it needs. Yet it’s the most important action to keep a KM solution running smoothly.  It suffers from three major problems:

  1. It’s costly. As a rule of thumb, costs of KM maintenance are roughly the costs of deploying the knowledgebase for the first year, plus a growth of around 8% average per year. Organizations don’t allocate nearly enough to start, nor do they budget for the growth in expense. 
  2. It’s always changing. An old professor of mine used to say that the value of knowledge decays at the rate of fifty percent per day. Knowledge (that applicability of content to specific situations based on multi-variate problems) is always changing as the variables that make the knowledge valid change as well.
  3. It’s very, very hard. The complexity of keeping up with knowledge is only mildly understood; knowledge validity is tied to the ability of knowledge and content to be the answer to any question. 

The underfunding of maintenance is the number one reason KM systems failed in the past. Not having updated knowledge available, knowledge being impossible to find, or not having knowledge when it should be there are the leading user complaints about KM systems.

Augmenting KM with AI is very useful to the above three areas, but as AI progresses towards machine learning (as AI is implemented and the system begins to learn) the true value of the augmentation comes down to solving the core issues of maintenance.

It would be very hard for AI to find a better discipline to augment than KM, not only because of the reasons cited above but because AI is about data, and content, and knowledge. Those are the focus of KM disciplines.

Read the next post in the series

Read the full White Paper here