How can you find the right procedure in just a few seconds within dense and confidential technical documentation, without ever leaving your secure environment? This is the challenge taken on by Airbus Defence & Space with the b<>com teams. Damien, project manager at Airbus Defence & Space, and Pierre Castel, project leader at b<>com, explain how this collaboration is transforming access to information.
A research program to deploy practical AI
Launched at the beginning of 2026 by b<>com, this collaborative research program focuses on enhanced efficiency. Built over two years with six industrial members, it is structured into 13 work packages. Its ambition: to deploy AI solutions serving the partners' professions and processes, and to transform still underutilized industrial data into real decision-making levers.
Among these work packages, the one conducted with Airbus Defence & Space (ADS) addresses a very concrete need: natural language search in confidential technical documentation.
What was Airbus Defence & Space's request?
Pierre, project leader at b<>com: "Their need was clear: to increase the efficiency of their technicians when searching confidential document databases. This raises issues of performance, but also of sovereignty, in a highly constrained environment.
We are therefore designing a system that operates entirely offline and allows interaction with a specialized language model. The challenge is to guarantee data security, control over processing, and confidentiality of usage, while providing a seamless experience.
To achieve this, we are combining several cutting-edge technological components: language models, text recognition (OCR), vision-language models, and multi-agent architectures. The goal is to build a tailor-made solution, not a generic product.
We proposed an iterative and pragmatic approach, with clear milestones and a pace suited to ADS's desire for speed. Each stage is validated with their teams."
Damien, what problem led you to launch this project?
Damien, project manager at Airbus Defence & Space: "We have very comprehensive and well-written user technical documentation, but its ergonomics no longer meet today's standards. In this documentation, an expert knows where to look, but a newcomer quickly gets lost, and the current search engines are not very effective. Since operators have little time to act on the system, we need to modernize this access. Moreover, we face a significant constraint: we cannot host these solutions on civilian infrastructures, and client-side options are limited. So, it's not about integrating an off-the-shelf solution, but adapting the technology to a highly constrained environment. We already knew about b<>com's AI expertise, so it was natural for us to turn to them."
What does b<>com's expertise bring to the project?
Damien: "Generally speaking, it's the combination of strong scientific expertise, rigorous and flexible project engineering, and a real ability to work within a secure framework.
From a technical standpoint, you quickly sense true technical expertise with strong ties to academia: we regularly interact with PhDs who advocate for a deep scientific approach. We're not just integrating existing, turnkey modules, but engaging in in-depth reflection on the questions we raise.
On the project side, you sense the agility of a start-up, real responsiveness, all while maintaining the level of management professionalism known at Airbus."
What impact will the solution have on your team and more broadly at Airbus?
Damien: "First of all, an obvious time saving. By making access to information easier, we can hope for greater responsiveness when breakdowns occur. Less directly, it facilitates skill development: by simplifying access to information, an operator will be more inclined to dive in.
Finally, this project boosts our team's skills in the field of AI, but also those of other teams that might have the same need. With b<>com, we demonstrate that the significant constraints mentioned earlier can be managed and can definitely be deployed in our client environments."
AI expertise that adapts to the most demanding sectors
Beyond this project, the collaboration illustrates a capability that b<>com offers its industrial partners: translating cutting-edge research into solutions tailored to the constraints of each sector, including the most sensitive. In defense, this requires top-level project engineering: a secure environment, end-to-end control, confidentiality and sovereignty requirements, from R&D to integration.
Our strength is combining complementary expertise to carry out a project from start to finish, from research to demonstrator, while integrating from the design phase the objectives of end users and the security constraints of their field.
