AI was impressive enough when skilled developers could use ChatGPT to troubleshoot code. But now, the tech has evolved to the point that workers with no programming background can use AI to build apps from scratch, refine them, and ship them, all in a matter of days or even hours.
Some folks call it vibe coding, but that undersells what's actually happening. There's a lot more than vibes behind these projects. The vibey part comes from how easy it is for someone to describe what they want to create in plain language and build fully operational software with AI in a single afternoon. Five years ago, "build me a dashboard that analyzes customer complaints" would have been a project request. Today, it's a prompt.
To learn more about the non-developers who are building software at work, we surveyed about 800 U.S. employees who have built or deployed tools with AI coding apps. We wanted to find out what exactly they're making—and how those newfound skills are changing their careers.
Key findings:
34% of people shipping software using AI tools have no formal programming background
54% of non-traditional builders say their tools aren't one-and-done—they're still in use today
20% of non-traditional builders using AI say this skill snagged them a promotion or raise
34% of people shipping software using AI tools have no formal programming background
Using AI for a pet project at home is one thing, but people are now creating high-stakes work projects with AI, even if they have little to no coding experience.
About 1 in 3 (34%) people building with AI tools at work don't have a formal programming background, meaning they didn't take college coursework for it and haven't worked as a developer or engineer before.
Non-traditional builders typically use the following:
82% use general-purpose chatbots, like ChatGPT or Claude
71% use AI coding assistants inside existing coding tools, like GitHub Copilot
40% use dedicated AI coding software like Cursor
The catch is that building something is now the easy part. Getting it to talk to the rest of your company's tools—without handing your credentials to an AI model—is where many non-traditional builders hit a wall.
Zapier MCP is designed for exactly that gap: it lets AI assistants like Claude and ChatGPT connect to 9,000+ apps while securely handling authentication, so the model never sees login details or API keys.
More than half of the people with no formal programming background create tools for customers or the general public
Generally, when people without a formal programming background build with AI, they gravitate toward basic apps with broad use cases. Most commonly, they're making:
Data analysis or visualization tools: 41%
Forms, surveys, or data collection tools: 39%
AI-powered chatbots or assistants: 37%
If you ever tried to knit a scarf the first time you pick up needles, you know it normally takes a while to create anything halfway decent without experience. But 30% of non-traditional builders go from initial idea (e.g., "I wish I had a tool to analyze these customer complaints") to a working tool, all in less than a day. (Somewhere, a project manager who survived a nine-month software rollout just felt a disturbance in the force.)
These aren't one-off projects, either. About half (49%) of the people with no formal programming background have shipped four or more tools in the last year.
And even though they're leaning toward simpler projects, more than half (56%) of non-traditional builders are making something customers or the general public will actually see. Only 44% are building tools just for internal use.
The number of these tools that end up in customers' or the public's hands means the stakes for getting the underlying infrastructure right are higher than most non-traditional builders probably realize. Zapier's governance layer handles that piece when you're connecting to other tools, so builders can focus on what they're actually making.
54% of non-traditional builders say their tools aren't one-and-done—they're still in use today
You might assume that any software written by someone who doesn't have extensive coding experience would fall apart immediately. Sometimes it does (my match-your-dog-with-a-celebrity app was deprecated real quick), but a lot of these tools are still serving their purpose.
More than half (54%) of non-traditional builders say most, or all, of the tools they've built are still used today. About a quarter (24%) say all of them are.
AI coding still presents challenges, of course. Nearly everyone with formal experience (90%) and without (92%) has run into a major challenge coding with AI at some point. Most commonly, non-traditional builders are running into code errors or bugs (38%), security and data privacy concerns (37%), and incorrect or unusable outputs (29%).
The security and data privacy issues are the most concerning—that's a PR nightmare you won't recover from. These are ops managers and finance leads shipping tools their whole teams use, often without IT's visibility into what those tools connect to or what they can access.
Technical builders will keep building; the data makes that clear. But the infrastructure beneath them needs to be able to keep up. That's why it's so important to build with governance and security in mind from the beginning. Zapier's governance layer bridges gaps in permissions, event logs, and controls over what agents can access without asking builders to become security engineers.
20% of non-traditional builders using AI say this skill snagged them a promotion or raise
Execs are pushing their teams to go all in on AI, and it turns out it benefits the employees, too.
One-fifth (20%) of non-traditional builders say their AI skills helped them land a promotion or raise recently. Aside from the literal cash, non-traditional builders claimed the following benefits too:
Feel more confident taking on challenges outside their job description: 39%
Have a competitive advantage over peers who aren't using the tools: 36%
Have opened up new career opportunities: 28%
Have received formal recognition for their work from their manager or leadership: 26%
The execs are still getting the benefits, though: 40% of non-traditional builders say their work using AI tools has helped them add more value to their organization.
And they're not stopping. Two-thirds (67%) of non-traditional builders say they plan to continue using AI tools in the next one to two years. And 36% expect to use them significantly more.
Looking to start building with AI tools? Here are Zapier's picks for the best vibe coding tools and the best AI coding tools.
You don't need to know how to code to build something, but you do need the right tools
My coding experience is mostly limited to editing my own MySpace page in middle school. But today, that's more experience than you need to ship something that will be used by a lot more people than your top eight friends. A third of the people shipping software right now don't come from technical backgrounds. They're just figuring it out as they go, and in some cases getting promoted for it. Good thing the infrastructure to do this safely exists now.
Zapier lets you build safely with AI, using OAuth-managed authentication for 9,000+ apps, so you can securely take action across your entire tech stack—directly from your AI tools. With security and governance handled under the hood, you're not trading speed for risk.
The bar for "who gets to build software" has dropped dramatically, and it's not going back up. The only question is what you're going to make.
Methodology
The survey was conducted by Centiment for Zapier. The survey was fielded between May 8, 2026, and May 11, 2026. The results are based on 818 completed surveys. To qualify, respondents were screened to be U.S. workers who have personally built or deployed a functional tool or application for work in the past 12 months using AI coding tools (such as Cursor, Claude Code, or GitHub Copilot). Data is unweighted, and the margin of error is approximately ±4% for the overall sample with a 95% confidence level.
Related reading:
