Developer Relations (DevRel) has always been about bridging the gap between developers and the technologies they use. Over the past few years, I’ve moved from one hackathon to another—as a hacker, a mentor, and now a DevRel. Even though my experience is still relatively short, I wanted to share some thoughts, as I believe we are witnessing a major shift in how developers interact with DevRel-produced content. To adapt to these changes, we need to evolve our ways of working.

One major trend I’ve observed in recent years, which has completely reshaped the way developers work, is the rise of AI-powered coding tools. 🧠

AI has fundamentally changed how developers approach coding. The traditional workflow—where developers manually write, debug, and refine code—is being replaced by AI-assisted coding environments. Tools like GitHub Copilot, ChatGPT, and Cursor are actively generating what looks like "production-ready" code with minimal effort from developers.

While this shift boosts productivity for experienced developers who can distinguish between good and bad AI-generated suggestions, it also introduces new challenges—especially for junior developers trying to learn coding from scratch.

Here's a few key ones...

Maintainers of open-source projects are feeling the impact of AI-driven contributions.

With AI lowering the barrier to entry for contributors, the number of contributors has increased. However, this has also led to an influx of low-quality pull requests. Many PRs require extensive refactoring, sometimes forcing maintainers to rewrite entire submissions. Instead of reducing their workload, AI has turned maintainers into teachers, spending more time fixing contributions than benefiting from them.

Looking at repositories on platforms like OnlyDust (which itself is a great tool) provides clear evidence of this issue.

In our ecosystem, this problem is further amplified by the incentives given to early contributors. Many projects reward developers with tokens or grants for contributing to the ecosystem’s growth. While I don’t think these incentives should be removed, I do believe that the trend of “farming” contributions or grants will only increase in the coming years.

With AI becoming the first point of contact for many developers, the role of documentation is evolving. Instead of reading extensive documentation, developers now use AI tools to generate code snippets or debug issues.

Even I, when working with third-party documentation—such as identifying which API endpoints to use for a specific feature—often just pass the link to an AI tool and wait for it to generate the correct request and parameters.

This raises an important question: Should documentation be optimized for AI consumption rather than human readers? If AI becomes the primary consumer of documentation, the way we structure and write technical content will need to evolve, like a new form of SEO.

Should we focus on AI-driven debugging tools or more sophisticated AI training models tailored for our specific product? We might see documentation evolve similarly to some Web2 platforms, featuring highly interactive playgrounds.

Documentation will still be necessary, but I believe we’ll shift toward more reference-style documentation rather than today’s detailed, context-heavy guides.

This also raises another question: Should we still invest time in creating lengthy tutorials and well-crafted written content for developers?

Everyone can draw their own conclusions, but in my opinion, content should evolve toward an on-demand approach. Instead of preemptively creating content, developers should first be encouraged to experiment with available tools (AI or otherwise) and then be guided toward specific content—or have content created for them when needed.

This shift would free up significant time for DevRels, leading me to my next point:

These optimizations will, in my opinion, allow DevRels (especially advocates) to focus on the one thing AI cannot replace: human interaction.

We need to increase our engagement with developers, guiding them through their development journey both online and in person. Hackathons serve as crucial touchpoints—but that’s a topic for another article. Community calls, direct conversations with projects, and initiatives aimed at building relationships with developers should be prioritized and strengthened. Jumping in your discord community voice chat and help devs with problem there can also be a good way to do it, you just need to make sure you have the right audience in this chat.

DevRels may also need to transition into teaching roles to better support junior developers who have never coded without AI assistance.

I teach a master course at a French engineering school, and I feel that this experience helps me better understand the phenomenon. There is still a lot of research and experimentation to be done in this area, but it’s an exciting challenge. Assessing the skills of developers students in the same way as before is no longer useful, we need to find new methods.

Everyone will have their own take on this evolution, and some may see things completely differently. As DevRels, our challenge is to navigate this shift and continue providing value in a world where AI handles much of the heavy lifting.

Whether through optimized documentation, smarter tooling, or strategic in-person engagements, the future of DevRel will be shaped by our ability to adapt to this new reality.

One thing remains clear: while AI can generate code, it cannot replace human connection. And that is where DevRel will always have the upper hand. ❤️

What are your thoughts on the future of DevRel with these changes in developer behavior? Feel free to share your ideas—I’d love to discuss this further on X/Twitter: https://x.com/Gwenole_M