It was 2019 when we started looking into how NLP and NLG would help us leverage the almost 'infinite' source of knowledge stored in thousands of historical documents and how the reuse of content would allow us to gain significant efficiency gains for our Regulatory and Safety functions. The question was how to leverage the promise of ML and the context in the SME's heads; the answer: network graphs and the concept of reusability paths. An approach that would allow us to bridge three challenges: 

1. How to ensure that we could extract the 'tribal knowledge' context of our experts 

2. How do we maximize value for the business in terms of speed, effort? 

3. How do we connect all those concepts and documents? 

In this presentation, we will show how the merge of a good data strategy, with semantic and structural mapping and an optimization algorithm helped us get to a first approach to maximizing value for an organization's know how.