Quickly onboarding new hires to become productive, expanding their knowledge, and addressing retention – these are all key areas where corporate knowledge management struggles with. This is mostly due to the vast amount of internal unstructured content that needs to be available to employees, customized and presented in an easy-to-consume way.

Traditionally employee onboarding and knowledge sharing were handled with specialized e-learning and training tools that require a lot of management, from data collection, structuring, creating curricula, workflows, and so on. For some large companies, this task is enormous – in terms of people, technology, and resources used – and rarely information is up-to-date. This is where Artificial Intelligence steps in.

An outdated approach to information consumption is that everything needs to be properly structured, organized and tagged in order to make it available for use. Structured content management projects are never-ending and resource-intensive. For some critical areas where this approach is still mandatory through regulation – like aviation or defense – this is and most likely will remain the standard. But other more agile industries, this approach has become an organizational burden.

The advancement of Artificial Intelligence, more specifically Text Analytics and Natural Language Processing, allows us to effectively make use of any unstructured data, in any language, through conversational systems that allows:

• Users to get access to data instantly through natural language conversations with IT systems/software.
• Data to organize itself with little human oversight, so that it instantly becomes accesible to relevant consumers.

AI Large Language Models (LLMs) based on the AI Transformer architecture can understand data in any format, in any language and from multiple repositories. The concept of making this data available to users – called Intelligent Knowledge Management – relies on powerful AI backend that can translate documents, classify them, extract named entities, concepts, compare them semantically and transform all written and image data into vectors that are then used in a Question and Answering scenario. Our implementation of this concept relies on our AI Factory technology and has been proven to work for virtually any scenario, from customer support, to Open Source Intelligence and employee knowledge sharing an onboarding. You can read more here: https://zettacloud.ai/use-cases/intelligent-knowlege-base/

Let’s take a practical example of a company from the Telecom industry that relies on a large sales force to expand their business across a certain territory. The services offered change frequently due to increasing competition; offers depend of network coverage, network expansion projects, government regulations, available equipment, reseller exclusive territories and discounts, and so on. All this information is stored in multiple repositories, some structured, some in raw scanned PDFs, emails, CRM notes and so on.

An Intelligent Knowledge Management system can programmatically access all these repositories through connectors (APIs or RPA bots), and then apply a pipeline of AI NLP algorithms to structure the data in order to Discover the key concepts and elements:

• Automatic Language Detection
• Automated Translation
• Automated Classification
• Extraction of Named Entities (name of people, companies, products, locations, …)

After the Discovery phase, the system will have all these materials prepared into a Vector Database for intelligent retrieval through semantic search aided by a Large Language Model, in order to use the metadata for focused filtering the most relevant information snippets, and prepare the user-machine conversation using also Generative Artificial Intelligence for providing the responses.

This approach will allow any employee to quickly understand the context of any task she/he must perform (the Discovery phase) and also be able to query any source of information and received relevant and exact answers from all data sources, in a natural conversational approach.

The added value of implementing AI to knowledge management across all industries is set to define a new standard in information management and its efficacy. As documented by an IDC research , the typical knowledge organization employing 1,000 knowledge workers wastes over $5.7 million annually searching for but not finding information, and 36% of a typical knowledge worker’s day is spent looking for and consolidating information spread across a variety of systems. These workers can find the information required to do their jobs only 56% of the time (Source: https://medium.com/work-bench/future-of-work-4-enterprise-knowledge-management-2-0-61ca08fd2a5d).