A field in an old system is useful to the person who already knows what it means. A new specialist can be hired into the building. The name of the field still belongs to someone who has watched a schedule slip, a supplier miss and a line stop. A licence does not teach that reading.
Airbus, the aerospace and defence company, had already put AI inside its products for more than a decade. The case study places that use in a small number of highly specialised technical roles, mainly around aircraft systems and performance. The wider opportunity sat in datasets the company already held, across supply chains, manufacturing, procurement, delivery scheduling and operational planning. Those datasets often lived in legacy systems. The case study says they could be interpreted by people with deep organisational and domain knowledge, which left external hiring as an incomplete way to get the work read. Airbus operates across aeronautics, defence, space and helicopters, and across divisions, countries and functions, in highly technical, operational, managerial and support roles.
Where the work already sat
The case study names two tasks at once. Airbus had to build advanced AI and data science capability at scale, and it had to make those skills accessible and relevant for employees and leaders outside the specialist roles. The approach it describes is long. It grew out of earlier digital work and was expanded over more than a decade. Capability sat inside workforce planning, competency frameworks and the organisation's learning systems. The case study presents that placement as the design.
The sequence started with data analytics and data governance. The case study describes a progression from data analytics to data science and then to AI. Once people were supported, they moved into new roles over time. The case study presents that internal move as the main path.
Pathways and mentors
Formal learning pathways were introduced for data analysts and data scientists, from beginner through to advanced. Around 2,000 employees completed them. The case study puts that figure at approximately 3 percent of the Airbus workforce. Those cohorts later became anchors, as mentors for wider learning and for the sharing of work.
Communities of practice sat beside the pathways. Practitioners shared projects, worked through problems and exchanged what had held up in use. The case study describes regular showcases, mentoring relationships and informal discussion as the way real examples became visible across the organisation. An internal podcast, AI for Newbies, explained concepts in short formats. The early cohorts were meant to be seen, so the work looked like something a colleague could demonstrate on a file the room recognised. The case study does not name the showcases, the episodes or the hour those mentors were given once the pathway had closed.
A competency on the role
AI competencies were written into the job and competency framework. Generative AI Essentials is named as one that could be added to any role, in any function. Assessments were used to find gaps and to connect those gaps to learning. The case study places this inside the annual workforce and competency planning cycle. Each year the Airbus Digital Academy produced an updated digital and data competence strategy, naming the capabilities the business needed next. HR business partners and line managers used that strategy when they recruited, redeployed and developed people. A course that never reaches the conversation about a role can be marked finished and still leave the job description as it was.
Two kinds of pathway are described. Essentials pathways were for employees and managers who needed a baseline understanding. Practitioner and specialist pathways were for advanced technical and analytical roles. People could add a competency beside the job they already had, without a new title. Some later applied for roles as data analysts or data scientists. Others changed how the current job was done and kept the title. The case study says many started using the skills partway through a pathway, so the work moved before any formal transfer. It does not say the calendar was cleared for that practice, or who accepted the new way of doing the job.
The formats were mixed on purpose. Formal courses, informal material, time in the community and application on real work sat together. People could pick a format that matched the role, their confidence and the way they preferred to learn. Data governance, ethical use and sustainability ran through the pathways, including the early levels.
The gate in front of the tool
Mandatory training was introduced before an employee could use generative tools. The case study describes that training as the condition of access. It does not say every pathway was compulsory, and it does not describe a consequence for a module left unfinished. Data governance was meant to make clear what could be shared with a system and what had to stay out. Responsible and sustainable use was treated as a starting capability, so that a trial of a tool sat inside protection, ethics and accountability.
The account also describes a form of recognition that a dashboard does not hold. Annual events for AI and analytics marked work that had been done. Senior leaders took part. Over time, the case study says, engagement became self-sustaining and needed little central coordination. The reason on the page for that continuation is visibility: showcases, mentors, a short podcast and leaders in the room when the work is recognised. The case study does not name who runs the community once that central coordination has faded, or what happens to a showcase when a mentor moves roles.
The hours in the pilot
The publication is plain about the status of the numbers. Outcomes in the case studies come from interviews with the organisations and from validation with them. They have not been independently audited or verified as performance data. Where a time saving is mentioned, it is the organisation's own reported evidence. A learning lead can hold the Airbus figures as Airbus's account.
On that account, uptake of the pathways, participation in the communities and completion of the mandatory training stayed high, with no evidence of a drop-off after the first deployment. The case study describes a move from scattered experiments to capability organised across the organisation. It places Airbus at what the publication calls the strategy stage of an AI skills adoption pathway, stage 8, where use lines up with organisational goals. It also says there is emerging evidence of a move toward a later scaling stage, through adoption across the organisation and a productivity effect that can be measured. The page does not publish the measure that would mark that later stage as reached.
A pilot of generative tools is the concrete hour. In a trial of approximately 2,000 employees, Airbus found that the average user saved up to four hours a week when an assistant was used effectively. The savings were seen mainly in office-based roles outside production, on routine tasks, analysis and administration. The case study ties those hours to the training and the data rules. It also lists effects that are harder to time: more confidence among people who are not in technical roles, stronger engagement across career stages, clearer problem framing and better decision quality, and a wider awareness of ethics, data and sustainability. Airbus, on this account, treats the hours as measurable and treats the longer value as capability that is hard to put into a simple financial figure.
The same account describes movement inside the company. As people gained the skills, many could move between roles and functions. The case study calls that an internal labour market: people stayed, and the capability spread. It does not say how many moved, which functions they left, or whether a team that lost a mentor was given another. The pathway cohort of around 2,000 and the pilot of approximately 2,000 are both on the page. The case study does not say they are the same people.
What this means for L&D
A learning lead can separate three decisions the case study has already made visible. The specialist pathways came first, at a scale the company can state, around 2,000 people and about 3 percent of the workforce, and those people were then used as mentors. The competency could be added to a role without a new title, and the Digital Academy's annual strategy was handed to HR partners and line managers for recruitment, redeployment and development. The generative tool waited. Training came before access, and the rule about what data could be shared travelled with it.
Three questions are still open, and they are the ones to ask before copying the shape. Who keeps the community, the showcase and the mentoring once the account says central coordination has faded. How the hours were produced, given that the publication calls them self-reported and unaudited, and given that the wording is up to four hours a week for the average user of an assistant used effectively. Whether the pathway cohort and the pilot cohort are the same 2,000, and whether anyone was given time once the competency sat on the role. A gate in front of a tool can be real. The hour, the owner and the overlap can still be blank.
Collective Campus has done capability building for teams since 2015. The AI training workshop is for a team that can bring a live task and is allowed to practise on it for long enough that the queue does not win the same afternoon. A single demonstration for every function fails for the reason set out in role-based AI training. What happens when the first week looks busy and the old work returns is in your AI rollout stalls in week two.
Count the people who finished a pathway other people can see, then look for the rule that sits in front of the tool. The case study records the headcount on the pathways, the competency that can be added to a role, the annual strategy in the hands of HR and line managers, and a self-reported saving in a pilot of about 2,000 office users. It does not record the owner of the community after the centre steps back, the method behind the hours, or whether the two cohorts are one group. The next decision is those three, made in the function that already holds the work.
Sources
Skills for AI: What works for AI upskilling in the UK, supporting case studies, GOV.UK, updated 27 July 2026. The Airbus case was developed through qualitative interviews and validated with the organisation. The publication states that reported outcomes have not been independently audited. GOV.UK.
