While we are waiting (or fearing) for completely autonomous systems to do all our work, a more feasible approach is appearing that focuses on augmentation rather than automation. One type of usage gaining ground in organizations’ around the world is the knowledge agent. A knowledge agent is an agentic system focused on a specific domain of knowledge. Through the knowledge agent the organization’s knowledge becomes accessible in a different way that provides new unique opportunities for the employees. Here are four of the most common ways knowledge agents boost employee capabilities.
Mentoring – If built correctly it can provide mentoring and on-the-job teaching for employees by referencing guidelines, Standard Operating Procedures and other training material. To achieve this effect the organization needs to be disciplined in its approach to process documentation. New employees would be able to learn by doing by asking the agent for step-by-step guidance rather than waiting for experienced colleagues’ availability.
Validation of thoughts – reviewing work products or sanity checks is another way to employ knowledge agents. This is done by focussing on a particular domain and orchestrating access to the organization’s documentation. An engineer might assess the feasibility of a design validated against experience from previous projects; it could also do an automated review rather than ask colleagues to take time to do it.
Sparring – sometimes you get stuck and need somebody to spar with on ways forward or ideas that you have. The knowledge agent can act as a peer that ideas and approaches can be tried out on and get immediate feedback. It just needs to have enough specific background knowledge about the domain. If you are working on a bid the knowledge agent may spar with you on win themes or how to structure the project based on your particular business model and strategy.
Information retrieval – being able to access sources across different repositories and modalities (images, video, sound) the knowledge agent opens for new ways of retrieving knowledge. If provided with sufficient constraints through specific semantic models, it will be able to navigate the organization’s knowledge base in a way no other solution can. It can find all projects where a specific approach or product was used.
Knowledge agents thus work not primarily by automation but by augmentation. Think of using a knowledge agent as a cybernetic add on. It extends and augments existing capabilities of the human, like an extra limb or X ray vision. It is not seamless to adopt for users since an extra limb and its utility might take some adaptation and getting used to. It will not always work exactly as you imagined and it will give the occasional completely off answer. That is to be expected by cybernetic adaptations; after all X-ray vision may not be convenient to use if you try to read the paper.
There is consequently a need to continuously use it in day-to-day work to understand where it provides value and how, and where it just does not help. You need to exercise the new cybernetic limb. Knowledge agents therefore still need to be carefully implemented in an organisation and embedded in the way people work to create value. There is a learning curve but the upside is the promise of becoming a cyborg with superhuman abilities.
Photo by DIEGO SÁNCHEZ on Unsplash