The Top AI Tool for Devs Isn’t GitHub Copilot, New Report Finds
For this week’s episode of The New Stack Agents, I talked to Scott Carey, the editor in chief of LeadDev, a platform that started as an events business that aims to help engineers become better at the management and leadership side of software engineering. LeadDev recently published its first AI Impact Report, which looked at how developers and the organizations they work for are using AI right now.
The report, which our research director Lawrence Hecht also worked on, provides a nice snapshot of where engineering orgs are in their AI journey right now. Some of the key findings were that while two-thirds of developers and engineering orgs are actively using AI tools and models, with another 20% in the proof-of-concept phase and 13% in a more exploratory phase, only 2% of respondents said they had no plans to use AI tools and models at all.
Image Credit: LeadDev.
“I want to talk to those [2% of] people, because I’m fascinated by where they work and why they’re not adopting these tools, because they really, really are the outliers here,” Carey said. “And then on the flip side, this is just now part of what people are doing.”
As for the tools people use, LeadDev’s report found that 43% of the respondents were using Cursor as a tool that was funded by their orgs (which was a bit of a surprise), with GitHub Copilot coming in second at 37%. The usual suspects, such as OpenAI, Google Gemini, Windsurf, and Anthropic’s Claude, comprised a middle group, followed by numerous tools with minimal share among respondents, including Amazon Q, Bedrock, Replit, and Lovable.
Image credit: LeadDev.
Given that the majority of respondents were professional developers and that the question was about tools funded by their organization, it makes sense that vibe-coding tools like Lovable were relatively low on the list, though I would’ve expected Amazon to fare a bit better here.
One thing that’s been interesting in the report is that the engineers who responded to it mostly used AI and large language models for things like code generation, meeting summaries, writing documentation and other content, and for researching and learning new concepts. They did not, however, typically use AI for data analysis (7%), testing and QA (7%), IT operations automation (3%) or to deploy their code (2%).
“The idea of AI DevOps, all these things that the vendors get really excited about, our respondents are not there yet. They haven’t worked their way down or through the software development lifecycle yet to really point it at those problems,” Carey said. “And I find that frustrating, because I think that’s where they have the most impact — or have the potential to have the most impact — but it’s not there and off-the-shelf, and it’s not where the vendors are focusing their attention.”
One interesting question that has vexed the industry for a while now is how effective developers themselves think these new AI tools are at making them more productive — and whether that’s actually the case. A recent report from METR showed that while developers think AI improved their productivity, these tools actually slowed them down by 19%.
LeadDev’s respondents, too, mostly believe that AI has made them more productive, with only 5% saying there was no change and 10% saying it was making them less productive. But 26% also said that they weren’t sure or didn’t know, which, as Carey noted, points to the fact that many organizations also don’t have a way for tracking developer productivity.
Image credit: LeadDev.
As these organizations adopt AI, another perennial question is how that will impact junior engineers, who are most likely to be replaced by AI tools. The majority (54%) of respondents to LeadDev’s survey believe that their organizations will hire fewer junior developers in the long term, and that their jobs will change to focus more on supervising agents.
For more of our discussion, which also included segments on how engineering organizations are spending on AI tools and how Carey, as a journalist, uses (and doesn’t use) AI in his own work, have a listen to the full episode or watch it on YouTube.
