80%

Reduction in SWMS creation time

 

$Y

Saved annually by reduction in manual labor

 

Z%

increase in consistent SWMS formatting and content

Problem Statement and Solution

An SWMS document is a regulatory requirement before starting any HRCW (High Risk Construction Work). The creation and submission of this document is mandatory by contractors & agencies involved in HRCW. Each of these documents outline the risks and mitigation strategies (among others) for an activity, and are very specific to the kind of work being done. The creation of this document by a contractor / user of the client’s platform required anywhere between 1-3 days of effort per document. Additionally, the quality of these documents were inconsistent due to different users having different skill levels and understanding of the overall requirements of an SWMS. We saw an opportunity to save a considerable amount of time as well as improve the quality of these SWMS documents, with the help of GenAI. The creation of the document also required advanced features such as user approval, partial re-write, edits, saves and others. 

 

We use agile framework of engagement to enable continuous improvement over the solution by taking into account the user feedback. It took 4 weeks for our client to realize the value of the solution we created for them.

 

Creating frictionless customer experience for Workplace Safety and Compliance Systems company requires faster communication and maintaining consistency in document quality. Our client, a leading organization in the field of workplace safety, faced a significant challenge in the creation of the SWMS. Their existing process for generating workplace safety documents, including hazard assessments, risk ratings, control measures, residual risks, and responsibility tables, was manual and time-consuming. This manual process not only resulted in inconsistent document quality but also led to formatting issues and extended document creation timelines. Our client recognized the need for a more efficient and error-free solution to meet their workplace safety documentation requirements.

High level objectives

• Reduction in end-to-end SWMS creation time

• Increase in formatting consistency of SWMS

• Reduction in manual efforts in SWMS creation

 

The solution involves the following components:

1. Web based UI designed in Streamlit which interacts with the backend engine using Rest APIs. The customer logs into the UI and uploads a job description. 

2. The Task List Generation Engine which uses OpenAI GPT 4 LLM generates the task list and provides that to the user for verification. 

3. Langchain is used as the Orchestration Engine where prompt template and guardrails components are tuned using Prompt Engineering to give the most accurate task lists.

4. Safety Table Generation Engine consumes the task list after user verification where safety tables, encompassing hazard assessments, risk ratings, control measures, residual risks, and responsibility tables are generated. The engine further generates a PDF file which can be downloaded from the UI for downstream consumption. 

5. Feedback Engine is built using the Chroma vector database which is used in the Task List Generation Engine to add context to LLM and continuously improves the engine performance.

6. For Deployment, Azure container app is used which in turn uses Kubernetes for scaling up as the demand increases