Scott & Ewing Success Story

Using Secure AI to Transform Tender Writing and Compliance

Scott & Ewing Group manages a high volume of tenders, PQQs, and supporting documentation across multiple projects and clients. These documents, primarily authored in Microsoft Word, are stored across numerous folders, making information retrieval slow and cumbersome. While the content needed to respond to bids is available, the manual search process places pressure on the bid team, risks inconsistencies, and limits the efficient reuse of high-quality responses.

The Challenge

Scott & Ewing Group’s challenge was not a lack of information, but how that information was stored and accessed. Key tender and company knowledge existed almost entirely in Microsoft Word documents spread across multiple folders and projects. Bid teams were required to manually search large volumes of content to answer routine but critical questions, increasing time pressure and the risk of inconsistency or unsupported claims. At the same time, Scott & Ewing needed to maintain a consistent professional tone, ensure factual accuracy, and meet strict data confidentiality requirements, ruling out the use of external, cloud-based AI tools. Any solution therefore had to operate solely on internal documents, provide traceable evidence for every response, and support human reviewers rather than replace them.

The Solution

A bespoke Retrieval-Augmented Generation (RAG) solution was designed specifically for Scott & Ewing’s tender process. Existing Word documents were automatically converted into searchable text and indexed using semantic models, enabling content to be retrieved based on meaning rather than keywords. A local large language model then generated responses grounded only in the retrieved evidence, ensuring accuracy and auditability. A custom prompt framework enforced Scott & Ewing’s tone of voice and required source referencing and reviewer notes, while a simple chat-style interface allowed users to query documents in plain English. The entire solution runs locally using Scott & Ewing’s own data, ensuring confidentiality and full control.

Yvonne Conway, Scott & Ewing

“The support we received helped us explore new ways of improving how our bid team works. Jade took time to understand our tender processes and explained what could have been a complex technical concept in a clear and practical way. The proposed Retrieval-Augmented Generation (RAG) approach demonstrates potential to reduce time spent searching documents, improve consistency, and support evidence-based responses, while maintaining strict confidentiality requirements. Importantly, the solution is designed to support our team rather than replace them. The process to date has been collaborative, straightforward, and aligned with our business needs.”

Project Impact

The RAG system has delivered significant benefits for Scott & Ewing’s bid process:

Instant search: Users can quickly retrieve information across all uploaded tender and company documents.
Draft generation: Produces draft tender and PQQ responses with cited evidence from source files.
Gap identification: Highlights missing information or areas requiring human verification.
Time savings: Reduces the time spent searching and rewriting similar responses.
Improved quality: Enhances consistency, accuracy, and compliance across submissions.
Supportive, not replacing: Provides clear evidence summaries and reviewer guidance rather than replacing human oversight.

This proof-of-concept demonstrates that generative AI can be safely and effectively applied to Scott & Ewing’s bid process, delivering measurable efficiency gains while maintaining control, accuracy, and professional standards.

Our Impact on UK Industry and Society