Active Knowledge Management at Technical Enterprises
I recently came across a post from a Knowledge Management executive at Fortune 50 company, a company with a heavy technical and discrete manufacturing footprint. The post included an info-graphic depicting the pillars of the company’s Knowledge Management (KM) strategy. I was a bit startled to see ‘Wikis’ and ‘Lessons Learned data base’ as two of the seven pillars and here’s why:
Technically intensive companies are becoming ever more model-centric.
PLM, CAD, CAE, BoM, Simulation, etc, are fundamental components of product creation and decision-making processes. As this trend continues, knowledge management as a strategy will be compelled to interface directly with these models. The center pillar of this Fortune 50’s KM strategy was ‘Communities’. While ‘Communities’ are, and should be, a foundational component to the KM process, KM's interface to the model-based enterprise represents the next opportunity to improve and assist in the collection, refinement, and application of knowledge. This interface will replace Wiki and Lessons Learned type of passive approaches with an active and in-the-flow-of-work use model.
If you currently use one or both of the legacy approaches, consider the following questions:
1. Can you create fast learning cycles that don’t rely entirely on brute force human effort ?
When KM interfaces directly with models, data about the use of knowledge can be gathered by linking together: a. specific context, b. the knowledge to be applied, and c. outcomes. This trifecta data-set is ripe fruit for AI. Think about how the navigation app on your mobile phone collects data about other drivers and applies AI to transform this data into value generating insights like traffic and ETA; map fidelity is improved, and end-user confidence in the navigation system increases (creating a virtuous cycle). In this nav app example, 'specific context' = destination/location, 'knowledge to be applied' = suggested turns expected travel times, and 'outcomes' = actual turns and actual time taken. Similarly, KM must turn model-based metadata about knowledge reuse into valuable insights and new learning.
2. Can knowledge be delivered when and where it is needed without the need for end-user search?
Think again of how the navigation app works on your cell phone. Do you get all the possible turns, or just the turn that is relevant to you based on where you are in your current journey? Searching through document repositories or lessons learned databases as a KM process is sub-optimal at its very best. While Top-down ontologies/taxonomies may be of some assistance, they are not the solution.
3. Can knowledge health be measured through analytical techniques and used in reducing noise in the knowledge portfolio?
One of the traditional failure modes of KM is ‘noise’. Noise is poor knowledge quality in the form of inaccuracies, incompleteness, verboseness, or lack of applicability to new contexts. Relying on brute force effort to manage noise in the KM process is better than turning a blind eye to the problem, however, the connection to the model-based world again provides the fertile data-set to detect and address noise analytically. A high signal-to-noise ratio in the knowledge base is a critical KM success factor. Imagine using your navigation app if you were only 70% confident it was going to take you to the right destination. You wouldn’t use it would you? Nor will an employee use a knowledge system with noisy content.
The fundamental enterprise need to capture and share explicit knowledge is real and this is why enterprises pursue Wikis and Lessons Learned repositories. However, it is now clear that an active, integrated, AI assisted approach that leverages the model-based enterprise will out perform these legacy approaches.
The International Knowledge Aware Association has been established as an independent organization with a goal to accelerate the growth of these important ideas. If you are interested, please join us: www.knowledge-aware.org