Valeo, a pillar of innovation in the automotive industry, is revolutionizing its production with artificial intelligence. In the face of an *exponential increase in the code* required for its embedded systems, the company is equipping itself with powerful tools such as Gemini. The reduction of development cycles and the acceleration of updates are becoming imperatives. This quest for efficiency focuses on optimizing human resources, without replacing the valuable expertise of its developers. The integration of artificial intelligence is thus transforming Valeo’s digital landscape, generating bold and tailored solutions to its contemporary challenges.
Increasing productivity with generative AI
Valeo, a major player in automotive equipment, is massively integrating generative artificial intelligence to optimize its code production. The growing demand for embedded software in vehicles necessitates an increase in the quantity and quality of code produced daily.
Cédric Merlin, Director of AI at Valeo, emphasized that the increase in software in cars, as well as their ability to benefit from continuous updates, has disrupted the industry. This growing pressure translates into an exponential need for lines of code. At the same time, development cycles have been reduced to two years, posing a challenge for code production.
A strategic collaboration with Gemini
The French equipment manufacturer has resolutely chosen Google’s Gemini as the main platform for its technological challenges. Valeo has equipped approximately 5,000 of its 9,000 systems and software engineers with Code Assist, a revolutionary code assistant. This strategy aims to increase developer efficiency, not to replace them.
The role of Code Assist is to enhance developer productivity by allowing them to focus on strategic tasks. The result is an increase in code production while maintaining a stable team, making artificial intelligence an indispensable complement for experts.
Reduction of development cycles
Valeo is deploying generative AI to simplify the entire software development process. Through semi-automated workflows, the company strengthens its processes, significantly reducing the time to market for products. AI agents notably act during the generation of unit tests and the detection of anomalies.
Cédric Merlin points out that AI is not only used to produce code but also to ensure its robustness. The automatic corrections proposed by AI represent a significant advancement. The developer has the option to accept or reject these proposals, thereby improving the validation process.
This model offers reduced development cycles and the capability to manage increasingly complex projects while adhering to security and quality standards.
Challenges faced with low-level code
Despite advancements, one challenge remains: the development of low-level code, essential in the automotive industry. Generative models, while effective on languages like Python, encounter limitations with specialized embedded code. This challenge requires a proactive approach to circumvent those limitations.
Valeo is working closely with Google to refine AI models, particularly Gemini, to adapt them to the specific requirements of automotive code. This fine-tuning allows for training AI on low-level code corpuses, thereby increasing its effectiveness.
Strategic partnerships and technological innovations
Valeo also benefits from a close partnership with Google Cloud. This connection allows them to access new technologies before they hit the market. Anticipating specific needs and adapting solutions to the imperatives of the company constitutes a significant competitive advantage.
This approach is part of a broader vision of integrating AI into various functions, aimed at increasing overall productivity. A synergy between automation and human expertise promises to redefine the contours of software development in the automotive industry.
Frequently Asked Questions about Valeo’s productivity increase through artificial intelligence
How does Valeo use generative AI to increase its code production?
Valeo leverages generative AI mainly through the Code Assist tool, based on Google’s Gemini, to enhance developer efficiency and generate a larger volume of code while respecting security standards.
What are the main benefits of adopting AI at Valeo?
The adoption of AI enables Valeo to shorten development cycles, produce more robust code, and adapt more quickly to the growing demands for complex code in the automotive industry.
What role does Gemini play in Valeo’s development strategy?
Gemini is the primary generative AI model used by Valeo for various use cases. It improves the productivity of engineers and automatically generates tests and corrections in the code.
How does Valeo manage low-level code development with AI?
Valeo faces challenges in low-level code development, essential in the automotive industry. In collaboration with Google, Valeo works on fine-tuning models to adapt Gemini to the specific needs of automotive code.
What tasks are now delegated to AI at Valeo?
AI at Valeo is used for generating unit tests, code review, and anomaly detection, allowing developers to focus on higher value-added tasks, such as design and validation.
Has Valeo observed a measurable impact of AI on its productivity?
Although Valeo does not share specific KPIs regarding productivity gains from AI, the company states that the use of AI has enabled faster development cycles and increased capacity to manage complex projects.
How does AI help maintain security and quality standards at Valeo?
By enhancing processes with semi-automated workflows, AI enables Valeo to produce code that meets security and quality standards, while facilitating the identification of potential errors through automatic interventions.
What other AI models does Valeo use alongside Gemini?
Valeo also utilizes models such as Claude, Llama, and Mistral for various domains, always selecting the most suitable model for each need from the range available on Vertex AI.
