Case Studies

Agentic AI Drafting of Preliminary Documentation

Technologies

NLP
LLM
RAG
AI

Client’s Need

Drafting the preliminary documentation of a new engineering development project is very often a low added-value activity. It requires a significant effort in terms of time and expert human resources, which are in large part squandered on repetitive and low-engagement tasks such as copying excerpts from legacy documents and checking adherence to corporate templates.

Approach

Development of a proprietary LLM trained on relevant legacy documents and fine-tuned to generate first-draft documentation according to corporate templates. The NLP solution is based on a RAG approach which speeds up considerably documentation drafting (and thus its associated costs) while giving full control over AI behaviour and document sources.

Main Outcomes

  • The initial draft is generated automatically in minutes.
  • The formal correctness of drafted documents is automatically ensured.
  • Senior engineering personnel can focus on high added-value activities related to the specifics of the new engineering project.