Description: Mapping and documenting several manual processes carried out in the field are challenging for large companies like Eletrobras, because they need to be standardized according to BPMN (Business Process Model and Notation) best practices. Due to the high demands, the complexity of these processes and the manual effort involved, we sought an innovative solution to improve this work.

The proposal is to use image-to-text and speech-to-text technologies, integrated with GenAI, to automate the mapping of processes and the construction of normative documents. The solution must capture audio, text and image information (with the inclusion of video being a differential), and automatically generate transcripts and workflow proposals or SOPs (Standard Operating Procedures), saving the team time and effort.

We are looking for a solution that automates data capture in the field and converts it into structured documentation, including process characterization, flowcharts and regulatory documents. The system must use language models (LLM) to accurately interpret and organize this data, mapping workflows based on verbal and visual descriptions, facilitating the creation of documents such as SOPs and other work processes.