Maggie Gao posted on X in June that she had analyzed more than 10 million lines of public data to build a free tool for New York City renters. The result, called Stoop, lets apartment hunters search any building across the city’s five boroughs for rent stabilization records, bedbug and pest reports, mold complaints and management company patterns before signing a lease. Gao built it using Lovable, a platform that turns plain-language prompts into working software without a line of hand-written code. She is one of a growing number of consumers who found a problem, could not find a product that solved it and built one themselves using AI.
Platforms including Lovable, Replit and Bolt from StackBlitz let users describe a product in plain language and get back working, exportable software, a practice known as vibe coding. More than 5 million people use Codex, OpenAI’s coding agent, every week, and more than 1 million of them use it for work outside software development, OpenAI reported. The barrier that once kept software creation inside engineering teams was never creativity or problem identification. It was the skill required to translate an idea into code. AI has removed that barrier.
Problems Consumers Are Solving Are Too Niche for Venture Capital
Enterprise software is built for the median user across a large market. The median NYC apartment hunter wants a clean search interface and good photos. The renter who wants to cross-reference a building’s violation history against its pest complaint record before signing a lease is a real person with a real need that no major platform has prioritized. Gao’s Stoop consolidates more than 10 million lines of public data into a single search interface, solving a problem at the intersection of city government data and consumer real estate.
The same dynamic is playing out well beyond real estate. Sarah Dooley founded AI-Empowered Mom after realizing that AI productivity techniques she had used in technology consulting could reduce what she calls the household’s mental load. “The mental load is really all of that invisible labor or ghost labor,” Dooley told PYMNTS. “The cognitive labor that has to happen to keep a household running.” Her AI assistant, Em, currently in beta, helps families plan meals, organize calendars, manage reminders and research purchases.
Parents are also using ChatGPT to write personalized bedtime stories and draft communications to schools, Good Morning America reported. Meta is piloting StoryKit, an AI app that turns a child’s photo and interests into a personalized story, in select countries, TechCrunch reported.
AI Dropped the Cost of Building
Small business owners are building invoicing tools and other back-office software without hiring a developer, a use case Lovable features prominently in its own marketing. In each case, the person building the tool is also the person who needs it.
The infrastructure that once separated a consumer with a problem from a business that solved it is getting thinner. Several tools built on Lovable and Replit by individual users have acquired users without a funding round, a formal launch or a product team. The cost of building a working product has fallen from months of engineering time to hours of prompting. A problem that affects ten thousand people is too small for a venture-backed startup. It is large enough for one person with an AI coding tool to address and monetize.
PYMNTS Intelligence found that 48% of online shoppers used AI during their most recent purchase across surveys in the U.S., Brazil and the UAE. The same shift is appearing on the supply side: consumers are using AI not just to find and evaluate products, but to build the tools they cannot find.
