ALTEN Group Case Studies Sharing
Methods Analytics, an ALTEN company based in UK, developed a conversational AI assistant, integrated with Retrieval Augmented Generation technology, to efficiently query company CVs and case studies, ensuring faster, more accurate identification of candidates and improving decision-making in the bid process.
Finding the right people can be as important as using the best technologies. When both are combined, the results can be transformative. Yet traditional methods of matching CVs and case studies with project requirements are time-consuming and prone to human bias. Methods Analytics – a company that provides end-to-end data services for healthcare, government and defense – leveraged AI to streamline the process of sourcing the most suitable personnel for their clients’ projects.
Challenge: To streamline the search for suitable employees and relevant case studies
Solution: An AI assistant tailored to provide bespoke results, save time and avoid human bias
Benefits:
- Seamless integration with existing systems
- Accessibility
- Flexibility and easy adaptability
- Time savings through automation
- Continuous improvement
- Streamlined project bidding
Key performance indicators:
- 30% reduction in production costs by embedding the CV Builder within the Skills Matrix
- Reduction in case study content creation time from 2 days to 1 hour, achieving a 1600% increase in efficiency
Offering the best
As a leading consultancy, Methods Analytics depends on the efficient sourcing of specialist personnel to stay at the top of their game, offering the best experts possible to their clients. To meet this challenge, they decided to leverage retrieval augmented generation (RAG) integrated with generative AI agents to develop a conversational AI assistant. The assistant queries the company’s database of CVs and case studies to provide suggestions tailored to meet each project’s requirements. It uses a custom retriever architecture to optimize document relevance and to ensure that it is easily accessible, it integrates with Microsoft Teams.
Tools and technologies
The key tools and technologies used to create this system included RAG and prompt engineering, focused on optimizing the retrieval process and refining prompt templates for concise, accurate responses. The retrievers were tailored to ensure that the document retrieval results would be highly relevant. By automating the updates of the RAG indexes on the Cloud, the system is able to reflect any changes in the CV and case study databases. The seamless integration with Microsoft Teams provides controlled access while maximizing reach.
Python was deployed for developing the core functionalities. Azure Services, including Azure Functions, Cognitive Search, Blob Storage, Logic Apps, and Container Registry, enabled cloud automation. GitHub was deployed for version control and collaboration, and Flask for building the web app components. The system is also easily scalable and highly flexible, adapting automatically to new data and requirements. Named entity recognition (NER) and optical character recognition (OCR) enhance document analysis and retrieval. Finally, the system is extremely user-friendly thanks to a conversational interface powered by OpenAI GPT-3.5 and Streamlit. Azure Bot Framework handles the chatbot functionalities.
Improved speed, efficiency, accuracy
The new system reduces the time required to identify suitable personnel, from one day to less than five minutes. Thanks to its improved accuracy, it allows for much more efficient identification of relevant skills and case studies, minimizing biases resulting from factors such as recency. Faster access to relevant case studies and employee profiles aids in more efficient bid preparation, reducing time and effort expended by the team. Integration with Microsoft Teams ensures that the solution is widely available while maintaining necessary access controls.