During the AI Action Summit in Paris, where President Emmanuel Macron announces an investment of 109 billion euros in AI for France, a critical question emerges: How can France and Europe truly compete with the United States and China in AI? While the need to train data and AI experts is clear — they build the essential LLMs and databases for AI applications — it is crucial not to overlook another key pillar: developers. These professionals transform technical breakthroughs into practical products and integrate AI components into scalable, working solutions.
Influential American investors like Andrew Chen predict that AI will become a commodity, increasingly accessible through APIs and open-source models. “Plug, Baby, Plug!” will apply not only to electricity and data centers but to AI models themselves. The real value will come from products that can leverage network effects and optimize distribution — which also requires developers and data experts, not just AI specialists.
No-Code tools and generative AI might suggest that coding is becoming less important. However, the AI startups experiencing rapid growth are not built without developers. Look at the latest Y Combinator batches: AI agent startups are multiplying, showing impressive performance (+10% average monthly growth and several million in ARR generated in less than a year).
These companies don’t just piece together existing tools: they develop unique AI solutions, relying on skilled development teams. No-Code is powerful, but in the AI era, its limitations are becoming clear, and coding remains essential.
Recently, I participated in a dinner with around fifteen entrepreneurs from the X-HEC Entrepreneur Master’s program. All were developing AI-centered projects. Most had initially considered developing their MVP entirely in No-Code, attracted by the speed and flexibility of these tools. But as they progressed, almost all adjusted their approach: while continuing to use No-Code for certain components (commercial websites, CRM, ERP), they prioritized code for essential product components, particularly their AI backend or industry-specific software.
This evolution is revealing: No-Code is an incredible accelerator for structuring a business, testing an idea, and automating internal tasks. But when it comes to building a differentiated product or a custom AI agent, its limitations become obvious: restricted personalization, complexity of integrating with advanced systems, scalability challenges, and an increasingly demanding learning curve.
Take another concrete example. Thomas and Paul, two former Le Wagon employees and seasoned entrepreneurs, are No-Code experts. After selling their No-Code training startup Emil to Le Wagon, they are now launching Akta Conseils, an AI startup specializing in administrative automation. Despite their No-Code expertise, they chose to partner with Dimitri and Julien, two former Le Wagon developers, to build their product. Why? Because they understand that a 100% No-Code approach would ultimately limit their potential, and code is essential for developing their AI agent.
Many entrepreneurs continue to learn to code and join Le Wagon’s web development bootcamp to launch their products, like John, who recently launched Jooc, an AI assistant for architecture agencies and engineering firms.
The rise of AI creates an interesting paradox for No-Code. While these tools allow easy integration of certain AI components, they also risk being supplanted by new No-Code tools that natively integrate AI. Massively investing in mastering a specific No-Code platform is exposing oneself to the risk of it becoming obsolete within a year. By comparison, acquiring fundamental web development skills represents a lasting investment that allows adapting to technological evolution without constantly having to relearn everything.
Another misconception is that No-Code is systematically simpler. In reality, the learning curve for No-Code tools is often underestimated. Interconnecting multiple No-Code tools can prove more complex than directly developing a custom solution. Entrepreneurs often end up spending more time working around limitations than truly building their product. Especially in 2025, coding with tools like Cursor allows multiplying productivity, which sometimes challenges the “Code versus No-Code” arbitration.
We often hear the question:
“Why learn to code in the era of No-Code and AI?”
But we could just as well ask:
“Why invest in potentially ephemeral No-Code tools facing AI? Wouldn’t it be better to consolidate solid technical foundations?”
There is no universal answer. Whether you choose code or No-Code, both approaches are valid, with the essential being to learn by creating and to move forward with full awareness of your choices.
No-Code remains relevant in many cases and will continue its rapid expansion. In digital marketing, for example, creating static sites or landing pages with Webflow proves more efficient than developing everything manually. Nevertheless, mastering Webflow becomes more intuitive when understanding HTML and CSS fundamentals, particularly for customizing a website.
For internal automation, solutions like Make, Notion, and Airtable allow developing personalized CRMs or ERPs and orchestrating sophisticated workflows. Here again, a technical understanding (databases, APIs, JSON) multiplies possibilities.
No-Code is therefore a precious toolbox. This is why we integrate certain No-Code tools in our Growth Marketing bootcamps or Skill Courses at Le Wagon (Skill Course Web Design or Automation). But beware of dream sellers promising to “earn millions with No-Code and AI without effort”. The reality is more nuanced.
The most accomplished No-Code experts often have a solid technical background. Shubham Sharma, one of the best No-Code experts in France, studied computer science at EPITA and followed Berkeley’s CS61 before working as a developer. Julien Mottet, an automation expert (Airtable, Make), trained in web development at Le Wagon in 2018.
Investing in code training in Europe does not oppose No-Code. On the contrary, these profiles will know how to judiciously exploit each tool at their disposal and make strategic decisions thanks to their technical understanding.
Today, successful AI applications are not just interfaces stuck to ChatGPT. They require a genuine technical architecture: adapted databases, optimized workflows, advanced integrations, intuitive user interfaces.
Developers play a key role. They integrate AI models into business applications, connect them to databases, manage APIs, and optimize performance to ensure scalability. A custom AI agent is not just a raw language model. One must structure data to avoid hallucinations, create intelligent workflows to transform responses into actionable insights, manage API calls, and implement caching strategies to optimize costs and speed.
A recent example perfectly illustrates the importance of this technical understanding: Cédric O, former State Secretary for Digital Affairs and co-founder of Mistral AI — a major French AI player — chose to learn coding at Le Wagon in 2024. This journey symbolizes the conviction that, even for technological innovation leaders, code mastery remains a fundamental asset.
This approach shows how essential coding remains, whether leading AI or developing tomorrow’s applications. Frameworks like Ruby on Rails and libraries like Tailwind allow developers to build robust products very quickly. AI-powered productivity tools further enhance developer capabilities. A coding assistant like Cursor enables developers to code 10 times faster, somewhat like having a senior developer by your side, capable of implementing new features very quickly. But one must always understand the code produced by this particularly powerful assistant.
The junior developer market is experiencing an adjustment period, but this reality affects all professions facing AI. AI and productivity tools allow experienced teams to accomplish more with reduced staff, whether in tech, marketing, finance, or legal sectors. The 2024 crisis forced many startups to slow down recruitment. However, massive AI investments announced in France and the USA could redistribute the cards in the coming months.
Despite this, coding skills remain more strategic than ever. If companies are hiring fewer juniors, it’s because they are looking for developers who are more versatile and able to fully exploit new tools. Look at the fastest-growing AI startups: none are being built without developers. AI does not replace engineers—it amplifies their potential. However, this power demands real expertise: coding is no longer just a ticket to a job but a fundamental skill to understand, optimize, and effectively integrate these new tools in a world where AI is radically transforming how we work.
Learning to code remains a particularly smart investment—not just for professional opportunities or entrepreneurship but also for gaining autonomy and adapting to market changes.
France is planning to create nine “AI Excellence Centers” and train 100,000 professionals by 2030. To support this ambition, Le Wagon is fully committed as a global leader in tech education. Our bootcamps train developers capable of building innovative AI solutions, data engineers and data analysts who master database creation and management for AI, as well as AI and data science experts, many of whom come from prestigious engineering schools and supplement their education with practical skills. With over 200 startups emerging from Le Wagon and more than a billion euros raised, mostly in France, Le Wagon demonstrates the tangible impact that tech education players can have—hopefully shaping the future development of the AI ecosystem in France and Europe.
Hi there! I’m Ben O’Mahony, Cytora’s People & Talent Director, my job is to bring
The role of Data Product Owners is to ensure value comes out of all the
“Being a FullStack Developer proficient in both Ruby and JavaScript has been a game-changer.” –
We’ve been working hard in London to encourage and support entrepreneurship and skills-based learning, through
“I want to make the leap from PowerPoint presentations to real working products by using
I’m the kind of person who’s always curious and doing a hundred things at a
