Build Erg'OH by Olivier Girard

Erg'OH — AI-Powered Ergonomics Knowledge Base

Business process automations using Retrieval-Augmented Generation to provide ergonomics and health suggestions. A proprietary AI-powered knowledge base interrogatable via API or conversational chat.

The Challenge

Erg’OH is an ergonomics consultancy whose expertise lives in a body of specialist knowledge that takes years to develop. The challenge was to make that expertise more scalable — enabling clients and partners to access accurate, context-aware ergonomics guidance without always requiring direct consultant time.

The solution needed to be trustworthy (ergonomics advice with safety implications must be accurate), accessible (API-callable by other systems as well as conversational), and genuinely useful as a standalone tool rather than a novelty.

The Approach

We designed and built a Retrieval-Augmented Generation (RAG) system around Erg’OH’s proprietary knowledge base. Key work included:

  • Structuring and ingesting the specialist ergonomics knowledge corpus in a format optimised for retrieval
  • Building the RAG pipeline combining embedding-based retrieval with generation via OpenAI and Gemini
  • Developing the API layer for programmatic access by third-party systems
  • Building the conversational chat interface for direct user interaction
  • Deploying the full stack on Google Cloud Platform with appropriate scaling and access control
  • Iterating on retrieval quality to ensure responses were accurate and appropriately caveated

The architecture was designed so the knowledge base can be updated by the client without requiring re-engineering of the system.

The Outcome

  • Proprietary AI knowledge base live and accessible via API and conversational interface
  • Ergonomics guidance available programmatically for integration into third-party workflows
  • Knowledge base architecture supports ongoing updates without system changes
  • Consultant time freed from routine information requests, focused on complex engagements

Technologies & Methods

Python, RAG architecture, vector embeddings, OpenAI API, Google Gemini, Google Cloud Platform (Cloud Run, Cloud Storage), REST API design, conversational UI.

About this engagement

This was a Build engagement: designing and building a production AI system around a specialist knowledge base, with both API and conversational interfaces. The focus throughout was on accuracy and practical utility rather than demo-ware.

Work on a similar challenge?

Book a 15-minute call to discuss your project.