Versent implemented a secure, enterprise-grade GenAI solution to transform its Request for Proposal (RFP) response process. The solution, named Bid Assistant, leverages Retrieval-Augmented Generation (RAG) to surface relevant internal content, enabling bid teams to work faster and more accurately. Designed with privacy, scalability, and usability in mind, Bid Assistant integrates seamlessly into existing workflows and delivers significant time and cost savings. The solution demonstrates how GenAI can be deployed in production at scale, following LLMOps practices to maximise business value.
Bid Assistant has transformed how we respond to business opportunities, saving time, boosting accuracy and empowering our team to focus on strategic, high-value work.
James Jackman Bids and Pursuits Manager
Challenge
Versent’s bid teams were consistently challenged by the volume and complexity of RFPs. Each response required a blend of technical precision and internal knowledge, often buried across disparate documents. The manual effort involved in sourcing, curating, and drafting content led to inefficiencies, duplicated work, and delays. Additionally, the team needed a solution that could securely retain and retrieve proprietary information and protect internal IP. The goal was to empower the team with a solution that reduced friction in the response process while maintaining high standards of accuracy and compliance.
Solution
The solution was architected using native Azure AI services such as Azure OpenAI, AI Search, AI Foundry to ensure modularity. It was designed to be cost-effective, scalable, and easy to deploy across different environments as Infrastructure as Code.
Overview of the Retrieval Augmented Generation pattern that powers Bid Assistant.
Throughout the build phase, the team adopted MLOps best practices to ensure the whole AI System Life Cycle was covered, including Data Preparation, Chunking, Automation and Evaluation. For example, during Data Preparation, the team dived deeply into the data, to choose an optimal chunking strategy. Before deployment, the system underwent rigorous evaluation to benchmark its domain expertise and establish a baseline for future improvements. This allowed us to produce an end-to-end solution that would be production ready.
Outcome
The deployment of Bid Assistant brought immediate, measurable improvements to Versent’s RFP response process. Bid teams could quickly locate and reuse key content across multiple proposals with far less manual effort. This not only reduced duplication but also enhanced the consistency and accuracy of responses. By leveraging a retrieval-augmented architecture, Bid Assistant empowered staff to work more independently, minimising reliance on technical subject matter experts and freeing them up to focus on strategic tasks.
Key Outcomes:
Annual Time Savings: By implementing Bid Assistant, we project an impressive annual time saving of 600 hours. This significant reduction in time spent on tasks allows our team to be more efficient and productive.
Cost Efficiency: Bid Assistant has delivered an ROI that is both immediate and substantial, ensuring that our investment is quickly recouped and continues to provide value over time.
Productivity Boost: The integration of Bid Assistant results in a significant productivity boost, enabling our team to focus on strategic, high-value tasks rather than repetitive manual processes. This shift not only enhances our overall productivity but also improves job satisfaction and team morale.
