Back in January, I launched a survey about the agentic coder landscape to understand how developers are really using AI coding tools. The survey closed at the end of March with 199 valid responses, and the results are now live.

View Results

Full transparency: I didn’t reach as many developer as I would have liked. Nonetheless, this was a fun project for me and I intend to do another version soon. Feel free to reach out if you have specific things you think should be asked in such a survey!

Highlights

Here are some of the findings that stood out to me:

• Prompt Engineers are the largest group. 31% of respondents fall at Level 3 on the agentic coder spectrum. Despite what social media may suggest, most developers aren’t running multi-agent orchestration workflows. Instead, they’re crafting prompts, reviewing generated code, and making targeted fixes.
• Claude Code is the most popular tool, used by 52% of respondents.
• Developers are mostly excited. 54% feel excited about the pace of AI adoption, 50% feel optimistic, and 48% feel curious. Only 20% of respondents see AI as a threat to their career.
• Social media is the #1 source of pressure.
• Cost is the biggest challenge. The top friction is the cost of AI tools, followed by lack of time to learn and experiment.

The full results include interactive charts, cross-tabulations, and methodology details. Go explore them:

Explore Full Results

Thanks to everyone who participated and helped share the survey. This is independent research driven by genuine curiosity — not funded by any vendor or affiliated with any company.