AI’s impact on software engineering continues, and more and more of that AI is packaged as agents as results from our newest pulse survey show agentic usage has almost doubled (59%) since we last asked about it in our annual Developer Survey. Companies are scrambling to provide agent harnesses, infrastructure, and applications, but we wanted to know whether people were actually using agents in their daily work.
Our latest pulse survey shows AI agent usage has nearly doubled since last year, jumping from 31% to 59%, but total agent takeover is not here just yet. While enterprise leaders are leaning into the operational perks and worrying less about costs, 63% of technologists still rarely or never let agents run entirely on autopilot. Instead, the industry is settling into practical, single-agent workflows where human review remains the gold standard. Leaders say the positive side of agents overshadows concerns about costs, accuracy, and security, while developers maintain that regardless of work quality improving, security and accuracy are still a major concern.
Here is a look at the data behind the adoption boom, the tools winning the race, and why industries like fintech are leading the charge. 1,100 developers and working professionals responded to our survey from late April and let us know their thoughts on AI agents.
Full autonomy is a risk agentic users are not willing to take. Most (60%) of survey respondents block agents from making unapproved system changes, and 68% prefer predictable, single-agent setups over complex, multi-agent configurations. The single agent workflow of choice for the majority of respondents (full-stack developers) is GitHub Copilot (65%) or Claude Code (50%).
Developers are still part of the process for coding, and they have been more skeptical about accuracy since we started asking about it in 2023. While there are fewer people using multiple specialized agents, let alone multiple orchestrated agents, the ones who do orchestrate multiple agents use agentic workflows daily more than single agent users. These daily, multi-agent users use Claude Code the most (70%), but also DIY with frameworks (OpenAI SDK 18%) and vector databases (ElasticSearch 17%).
Workplace agent usage has nearly doubled since last year, but it’s not just developers using them. When we asked about AI agents last year, we saw many developers were using AI. Fewer developers were using agents, but were planning to. This new survey shows 59% use agents at work at any frequency, compared to 31% in the 2025 Developer Survey. The growth is primarily in daily use, showing how these embedded tools have seamlessly become part of many aspects of work. That daily usage is partly driven by developers (40% report daily use), but also architects (52% daily use) and senior executives (50% daily use).
Executive sponsorship is actively shaping the modern enterprise tech stack. The differing results we see with students (38% daily use) and academic researchers (28% daily use) would suggest that the absence of typical workplace productivity goals and possibly the continued concern over accuracy in AI output remains a blocker for those in learning or research-focused settings.
Considering the growth in agents at work, and where the fire is being lit to encourage more daily agent usage, we also asked how everyone is feeling about the typical AI concerns. Our top five job role respondents let us know that the concerns are still there but diminish a bit for daily users. Executives and engineering managers are the least worried about cost: 75% of executives and 65% of engineering managers disagree or strongly disagree that cost is a barrier to using agents. In last year’s Developer Survey, 53% of users saw cost as a barrier to using agents. That’s fallen to 38% who agree or strongly agree that cost is a barrier.
Accuracy and security remain the top two concerns with using agents at work, both in this pulse survey and in last year’s Developer Survey. However, accuracy concerns have decreased from 57% strongly agreeing that accuracy was a barrier for agentic use to 47%. The percentage of respondents who strongly agree that security concerns are a barrier to adoption dropped to 44% from 56%. Architects and students are more sensitive to accuracy and security concerns, while engineering managers are less concerned.
Industries are taking up agentic tool use differently—this is reflected in some of the no-code tools we see growing in usage. Fintech and media/advertising industries are leading the way in daily agentic usage (55% and 50% respectively), even more than software development (44% daily usage). Fintech is having an interesting moment with the convergence of prediction market popularity, cryptocurrency entering major financial institutions, and sports gambling apps that all rely on data to power their real-time features. Agents and agentic tools are helping fintech continue to power these data products, whether directly through building internally or indirectly through providing data for the public to consume into agents of their own.
Media and advertising is having a moment with AI as well. Meta announced their AI advertising product in 2025 right as new no-code/vibe coding agents were starting to hit the market. Lovable and Base44 offered many a glimpse into how agents would impact traditional website and video marketing by offering an easy way to produce online assets in record time. Agentic no-code tooling is bringing more non-native technology users into the agentic development process. Among the no-code agent tools we asked about in the pulse survey, the most popular options were Lovable (28%), Replit (27%), and v0 (20%). All three of these tools have experienced growth in usage since the 2025 Developer Survey (Lovable and Replit both grew in usage by 22 percentage points, v0 by 11 points) and show high interest for usage in the next six months (Replit being the highest overall with 25%). Notable mentions for high interest also goes to Base44 (13%), Langflow (13%), and Glue (12%).
Preference for agent tools overall goes to coding tools that have been around since agentic AI started to become more established in late 2024/early 2025:
Agents may be the future of technology (or not). While fewer people see cost as a barrier, it remains a major factor with agents for the time being. Anthropic just updated their policy for subscribers who run up inference costs with agentic assistants like Openclaw. GitHub Copilot also announced new subscriber usage costs on top of subscription costs. Ramp continues to show the data doesn’t lie when it comes to corporate AI spend.
We are excited to bring all the insights together again for this year's annual Developer Survey. Rather than opening it in May this year, we are waiting a few weeks to get more of the growing community of technologists involved. While we wait, we have opened up an open-source GitHub repo for last year’s survey (with every year of Developer Survey data) for anyone who wants to get deep into the questions and submit feedback for this year. For those who want to know more about the Developer Survey, make sure to follow our blog where we will post all the latest updates.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。