Implementing AI solutions at a large automotive company

Implementing AI solutions at a large automotive company

Industry: Automotive

Organisation size
1,000 employees
Duration
3 months
People involved
40 people

Objectives and executive summary

The aim of the project was to introduce AI into the day-to-day work of the logistics, finance and project management functions of a large automotive company and to improve operational efficiency. The company had already decided to roll out Anthropic Claude and the licences were in place, yet usage had not taken off.

We started the rollout by assessing the tasks. We analysed around 70 use cases, roughly half of which proved feasible without system integrations, so the rollout did not have to wait for IT development. We built 10-15 use cases together with the users, and a 15-strong power user group is now developing the rest independently. All 40 employees involved use the solutions developed, and according to the employees’ self-reported measurement, close to 30% of capacity was freed up.

Client background and context

Our client is a large automotive company employing more than 1,000 people, where a total of 40 employees from the logistics, finance and project management teams took part in the programme.

The company had previously decided to roll out Anthropic Claude, and the licences were in place. However, usage did not deliver the expected results: the users concerned had no AI experience, and it was unclear which of their tasks the tool could help with and what data they could use in it. There were individual attempts, but these tended to increase distrust of the new tool. The organisation’s digital maturity could be described as low, which later created several challenges in carrying out the work.

Objectives and tasks

The task was to launch the practical use of AI, prepare employees and develop solutions that create business value, building on the existing licences, actual work tasks and the technical options currently available.

The main tasks of the programme were as follows:

  • Building the foundational knowledge and security awareness needed to use AI.
  • Identifying, assessing the business value of and prioritising role-specific use cases.
  • Developing the AI strategy and implementation roadmap for the area.
  • Developing the security rules and the governance framework for AI use.
  • Building the selected solutions together and developing internal knowledge that enables the team to continue development independently.
  • Measuring results and defining further development directions.

Challenges and difficulties

  • On several occasions we encountered resistance among users towards using AI solutions
  • The existing AI security policy was a general description that contained no specific data categories, conditions of use or guidance
  • The data needed for the work was available in several separate systems
  • Due to capacity constraints, the global IT function did not permit system integrations to be set up

Implementation

Common foundations

We launched the programme with a one-day AI fundamentals training that also covered the basics of AI security. Besides covering general LLM knowledge, security basics and the most important core features, the training also aimed to ease resistance.

Not every problem is an AI problem

We started the work by assessing the tasks. We collected descriptions of the typical tasks of the roles involved in advance and sorted them into three categories: tasks that can be supported by AI, tasks that can be improved through process automation, and tasks that require no IT solution and can be addressed through operational or organisational change.

Building on this, we held one-day use case discovery workshops, as a result of which we analysed around 70 use cases by business value and return on investment. Roughly half of these proved simple to implement without system integrations, so the rollout did not have to wait for IT development.

Strategy and roadmap

From the results of the use case and leadership workshops, we developed the area’s AI strategy (vision, objectives and focus areas) and the implementation roadmap for the identified use cases. The first months focused on use cases that were simple to implement yet strongly supported individual or team productivity.

Security and governance

We developed the AI security policy, including data categorisation, and the governance framework. The approved content was handed over to the key users through training; they were later responsible for delivering internal training and creating digital learning materials.

Building together

We built 10-15 of the quick-to-implement use cases together with the users in power user workshops. This was enough for the team to develop further solutions independently, based on these examples. The project managers received AI training tailored to the organisation’s own project management process, in which we covered the AI support options across the entire project lifecycle.

Results

All 40 employees involved use the solutions developed, even though none of them had any AI experience when the project started. A 15-strong power user group has been formed that independently builds new solutions and fine-tunes existing ones.

Before the rollout and after one month of active use, the employees themselves measured the time spent on their own tasks. Based on their self-reporting, close to 30% of capacity was freed up. On this basis, we made recommendations for redesigning roles and workflows.

The company plans to extend the use cases developed, and based on the experience gained it will decide on system integrations, automation-based AI agents and the related operational changes.