The possibilities are vibe coded

Thanks to AI tools, the barriers to who can now be a “tech founder” are breaking down. Vibe coding — which allows users to build software, apps, and websites via high-level descriptions of a project’s vision and intricacies — seems to be introducing a new class of founders. 

Now, experts across fields who’ve witnessed firsthand which problems are worth solving in their own domain can prompt and vibe code their way to tangible, digital prototypes. And the speed of AI can turn those ideas into executed products in days or, in some cases, hours. V1 pilots, it seems, are becoming the new napkin sketch.

On the secure vibe coding tool v0 by Vercel, which touts more than four million users, 63 percent of those users are non-developers, the company said in a 2025 report. The explosive growth of vibe coding apps like Lovable — which has roughly eight million users and as of June 2026, reports more than one million new projects built per week after less than three years in business — also shows how deep the interest in vibe coding goes. Lovable says four out of five users self-report as working in non-technical roles, and, in addition to tech, come from industries including education, retail, health care, and real estate.

“We’re moving away from the world where technical founders are the center of attention,” said Vitalii Dodonov, co-founder of Stan, a digital platform helping social media creators monetize their businesses. “The best people to build for humans are humans who understand those problems very deeply.”

Dodonov co-founded Stan in 2020 with John Hu, whose background as a creator provided the domain expertise to know what needed solving in the industry. While they built the platform via traditional software engineering, in 2025, the pair holed up in an apartment for two weeks and completely vibe coded Stanley, an AI agent designed to increase LinkedIn engagement for creators. Stanley, which launched in March 2026, now serves as a core business offering that the founders say grew to $3 million in annualized recurring revenue in a few months. 

“On the day of the [Stanley] launch,” Dodonov said, “250 people signed up for it, committing $100,000 in recurring payments for a tool that I haven’t written a single line of code for.” 

As the barrier to software creation dissolves into using natural language to communicate with vibe coding tools, deep domain expertise — and the language that comes with it — can become the ultimate tool and differentiator.

“The imperative to learn the domain language is about getting the best possible results from the LLM as you work with it,” said Mike Swift, co-founder and CEO of developer community and hackathon giant Major League Hacking (MLH). The community of more than five million developers includes students, early-career developers, and seasoned engineers.

Swift is a traditionally educated computer scientist, but his work hosting hackathons at universities and companies gives him a wide lens on the types of builders in today’s software ecosystem. He often tells a story of Andrew Ng, founder of DeepLearning.AI, who’s widely considered the most prominent authority in machine learning education. Ng and an art history professional took to AI image tool Midjourney to see who could generate better pictures. Despite Ng’s deep expertise in prompting, it was the art history professional who came out on top.

What Swift took from this example is that building something exceptional is becoming less about employing the best engineer and more about articulating what you’re trying to create in the language of an expert. 

Swift also noted that he has seen an expansion of founder types in his audience, which used to be mostly computer scientists or technology-adjacent staff. “[We’re] starting to expand into an even bigger pool of people who would not have considered themselves in our audience before,” he said. 

In April 2026, 10% of MLH community members were founders, double the year prior and the highest rate in seven years. And project finish rates at hackathons have doubled from about 20% to 40% per event, a trend Swift said stems from vibe coding. “It’s easier than ever for people to do more with less, whether it’s less people, less money, less time,” he said. 

Swift has seen an even bigger focus on the core business problems of customer value, user experience, and product-market fit. This has always been a major part of founding any company, but non-technical founders using vibe coding have more space to hone in on it from the outset, helping them differentiate their product in an increasingly saturated market. He believes ideas and taste now hold a heavier weight in the process.

Jessica Sophia Wong, CEO and founder of global venture network Yorkseed, is a non-technical founder who’s expanded her company through vibe coding. Before Yorkseed, Wong worked in interior design and digital marketing. Always ideating and managing relationships, her company began as a networking spreadsheet that went viral in Austin during the 2023 South by Southwest conference.

Wong launched Yorkseed as a networking ecosystem connecting founders and investors, launching summits at New York Tech Week and Toronto’s Collision Conference. It’s since grown to include hubs in places like Miami, Dubai, and Hong Kong, as well as gatherings at the United Nations, the World Economic Forum’s Davos Summit, and more. 

The community now has more than 35,000 people across its newsletter, events, and social channels, but Wong wanted to offer a software product to add value. She found friction in aligning her vision with potential technical partners so, in 2026, began vibe coding a software layer to serve as the crux of her offerings. The newly launched Yorkseed Connect features digital networking tools for founder-investor matching, deal flow tracking, and venture intelligence.

“Instead of spending all my time explaining what I wanted to build, I could directly participate in building it,” said Wong. Her work observing real-world problems gave her the foundation she needed to build on her own solution.

Wong mostly used OpenAI’s ChatGPT and Anthropic’s Claude Code to create Connect’s due‑diligence intelligence system for evaluating funds, deal memo generator, and cross‑platform messaging layer that syncs Yorkseed Connect with Telegram. There are also security features like two-factor authentication, though Wong said the platform doesn’t currently process sensitive information. But to plan ahead for future scaling potential, she ensured things were set up for engineers from the outset by organizing her entire codebase on GitHub. 

“Even though you’re vibe coding, you still need to have a little bit of an understanding of how certain things work,” she said.

Wong’s hardest lesson was operational. When she started vibe coding, she had recently purchased a new M2 MacBook Air. Had she known she’d be deep in development, she would have opted for a computer with more RAM. But other than that, the only thing she’d do differently is start building sooner.

Another first-time tech founder is Los Angeles-based fine art photographer Sage Causie, a creative building for other creatives. A recipient of grant funding from the Andy Warhol Foundation and Nikon, Causie has no technical background, but the experience in her field enabled her to vibe code Museboard, a free, no-login mood boarding tool.

“As a photographer, I needed a simple way to build and share mood boards without requiring clients to log into an app,” said Causie, who used ChatGPT to build out the first version of the product but refined it with Claude Code. After an eight-hour overnight vibe coding session, Causie launched Museboard in October 2025. To monetize it alongside her photography business, she’s building things like digital sticker packs in partnership with graphic designers that people can own and use on their mood boards.

Having the taste and domain expertise to successfully vibe code a great product is only half the battle. The next, often harder hurdle is actually getting that product into the hands of users amid a democratized startup market where customers have more options than ever. 

“AI can collapse the distance between idea and product, but turning a product into a real business still takes judgment,” said Elena Verna, head of growth at Lovable. “Understanding users, testing assumptions, building trust, refining the experience, and learning from real usage are business problems, not just coding loops.” 

While vibe coding has allowed many non-technical founders to hit the gas on building their products, some experts are concerned about what it means to have so many tools built by those without true technical know-how. They’ve flagged issues from security vulnerabilities to architecture failures that cause apps to falter or break entirely.

Case in point: AI-powered code vulnerability detector Veracode tested over 100 LLMs and found they opted for insecure coding methods 45% of the time. “That’s not an edge case, but a baseline problem,” said Stephen Hilt, principal threat researcher at AI security platform TrendAI. He said this unvalidated trust in AI is the biggest risk for vibe coders.

For the record, Hilt thinks software becoming democratized is a positive thing — but more people building doesn’t mean more people thinking about security. Threat actors know this and are already targeting vibe coded applications through methods like slopsquatting, where AI coding agents hallucinate non-existent package names, and bad actors register malicious software packages under these names, causing users to invite malware in their system. 

“What vibe coding introduces are errors of omission,” said Hilt. “It’s not like anybody did anything wrong. It’s more that nobody [technical] reviewed what the AI did.” 

Dodonov underscores that whether vibe coding is the right fit depends on what you’re trying to create and for how many people. “If you are responsible for processing [sensitive] data, you can’t be ‘vibey’ about it,” he said. “You need to know exactly what’s going on.” While new apps and products can still hail from a prompted vision, it’s important to be discerning about whether more thorough technical reviews ought to follow suit.

Yorkseed Connect, Stan, and Museboard are just some of the napkin sketches brought to life by vibe coding. While best practices of secure, functional development continue to ring true, using natural language and industry acumen to get from point A to point B has hatched a new class of founders, ones with the potential to be as successful in their pursuits as the technical founders we’ve become familiar with.