How CamyPro is helping to surface critical data in the defence supply chain

CamyPro builds quoting and workflow automation tools for manufacturing businesses. Its open-core platform combines automation, AI and deep expertise in manufacturability to help companies turn engineering designs into accurate production quotes faster than conventional methods allow. The company's existing clients use it to streamline the journey from design to production, cutting the manual effort out of a process that has traditionally been slow and error prone. 

That focus on extracting useful intelligence from complex engineering data is why CamyPro identified the opportunity of the Defence Testbed Accelerator as a natural fit.

The opportunity

The Ministry of Defence (MoD) holds around 1.3 million parts. Many are ageing, and some face obsolescence – meaning the original manufacturers may no longer produce them or the tooling to make them no longer exists. Additive manufacturing (AM), commonly known as 3D printing, offers a practical route around that problem: if a part can be digitally assessed and printed on demand, the MoD can reduce its dependence on traditional supply chains for components that are increasingly hard to source. 

The difficulty is that the information needed to assess whether a part is suitable for 3D printing is not stored in a database field. It is buried inside unstructured documents such as engineering drawings, material datasheets and design documentation. Retrieving it accurately, at scale, and in environments where security requirements are stringent, is a problem that conventional data tools are not designed to solve. 

Why the company joined the accelerator programme

The Defence Testbed Accelerator was delivered by Digital Catapult as part of the Made Smarter | Digital Supply Chain Hub, alongside the Ministry of Defence, the National Composites Centre (NCC) and the Manufacturing Technology Centre (MTC). The programme brought together five UK-based technology companies, working across two connected challenges, to develop and validate the infrastructure needed to support secure, distributed defence manufacturing. 

Challenge A focused on building connector-layer software that links to different original equipment manufacturer (OEM) product lifecycle management (PLM) systems within the testbed, and Challenge B focused on developing a federated digital inventory that aggregates standardised metadata from multiple OEM connectors.

CamyPro came with a problem statement clearly articulated, and the programme aligned with what the company already knew how to do – make complex engineering data accessible and actionable. The accelerator offered the opportunity to apply that capability in a defence context, working directly with MoD-aligned organisations and validating the technology in an environment where the stakes are real.  

The solution

CamyPro worked on Challenge A: developing the data extraction and normalisation infrastructure needed to populate a federated digital inventory of technical data pack (TDP) information. Its approach centred on a secure, event-driven platform built in five layers. The system pulls data from product lifecycle management (PLM) systems in the testbed, PTC Windchill and Autodesk Vault, queues it for processing, and then runs separate analysis pipelines on CAD models and engineering drawings. A locally deployed large language model (LLM) then consolidates and enriches the output, calculating high-level manufacturing properties including 3D printability assessments and qualification summaries. The processed data is made available downstream via a secure API using modern authentication standards.

The decision to use a locally run LLM, rather than a cloud-based one, was driven by MoD security requirements. Defence data cannot leave a controlled environment, so the AI processing had to run in an air-gapped configuration, isolated from external networks. CamyPro resolved this by selecting an open-weight model capable of running on standard CPU hardware without specialist processors, and built an orchestration and caching layer to keep processing efficient within those constraints. 

The results validated the approach. The system achieved a 98% consistency rate for categorical outputs using the locally deployed model, confirming that high-accuracy AI extraction is achievable within the security requirements the MoD demands.

Outcomes & results

The programme’s results met all four of CamyPro's primary objectives: secure PLM connections, metadata extraction, data normalisation, and a working TDP API. It also produced a clear application that goes beyond populating a digital inventory. 

By extracting and structuring the engineering metadata locked inside TDPs, the system can assess whether individual parts are viable candidates for additive manufacturing. For a customer managing 1.3 million parts, many of them ageing or approaching obsolescence, that is a significant capability. The MoD could use the output not just to build a digital inventory, but to actively screen its parts catalogue against AM viability. In turn, it could then identify candidates for on-demand printing before they become a supply chain problem.

The programme also expanded what CamyPro can offer commercially. The AI pipeline developed in the testbed has enhanced the company's existing quoting platform and created the foundation for a new product line – air-gapped generative AI workflows for analysing sensitive engineering data. That capability has already attracted interest from OEMs operating in similarly security-conscious environments. 

Customer perspective

CamyPro's experience of the programme was positive, with the problem statement clearly communicated, and day-to-day communication from the Digital Catapult team was prompt and responsive throughout. One-to-one sessions provided useful feedback at key stages of development. 

What the programme confirmed, for CamyPro, is a principle that holds across manufacturing contexts: the most valuable information is rarely the most accessible. The ability to retrieve it reliably, at scale, and within the security constraints of defence is where the real opportunity lies

“This programme showed how much value can be unlocked when the right problem, partners, and environment come together. It gave us a real-world setting to prove that critical engineering data can be securely extracted from unstructured TDPs and turned into actionable insight using locally deployed AI. We’re excited to build on this work, unlocking engineering data for future customers and enabling faster, more informed manufacturing decisions.” 

Adam Lofts, CEO, CamyPro