Consumer AI coding tools turn niche problems into working apps
AI coding platforms are letting non-developers build tools for renters, parents and small businesses from plain-language prompts.
By Rafael Ortiz · Fintech Correspondent
· 3 min read
Consumer AI coding tools are pushing software creation beyond engineering teams, with users turning narrow, personal problems into working products. Maggie Gao said in a June post on X that she processed more than 10 million lines of public records to create Stoop, a free tool for New York City renters, while OpenAI has reported that Codex draws more than 5 million weekly users, including more than 1 million using it for work outside software development.
Stoop lets renters search buildings across the five boroughs for records tied to rent stabilization, bedbugs, pests, mold complaints and landlord or management-company patterns before they sign a lease. Gao built the service with Lovable, a platform that converts plain-language instructions into software without requiring users to write code manually.
The wider shift is being driven by tools such as Lovable, Replit and Bolt from StackBlitz, which allow users to describe the product they want and receive functioning, exportable software. The practice is often called vibe coding, a term used for building applications through iterative prompts rather than conventional hand coding.
What is vibe coding?
Vibe coding is a way of creating software by telling an AI system what an app should do, then refining the output through further instructions. The AI handles much of the code generation, which lowers the technical threshold for people who understand a problem but have not trained as software developers.
That change is widening the pool of potential software makers. A renter can combine public housing data into a building-search tool, a parent can make a household assistant, and a small-business owner can create back-office software without first hiring a developer or joining a product team.
PYMNTS reported that Sarah Dooley founded AI-Empowered Mom after applying AI productivity methods from technology consulting to household management. Dooley told PYMNTS that the “mental load” covers the invisible cognitive work needed to keep a household running. Her beta assistant, Em, is designed to help families with meal planning, calendars, reminders and purchase research.
Other consumer uses remain closer to everyday content creation. Good Morning America reported that parents are using ChatGPT to draft school communications and generate personalized bedtime stories. TechCrunch reported that Meta is testing StoryKit, an AI app that creates a story from a child’s photo and interests, in selected countries.
Why are consumers building their own software?
Many of the problems being addressed are too specific for large software companies to prioritize, yet common enough to matter to a defined group of users. A mainstream apartment platform may focus on listings, photos and search filters, while a renter worried about building conditions may want public complaint and violation data in one place.
The economics of building have also changed. Lovable promotes invoice generators and other small-business tools as use cases for its platform, and PYMNTS reported that individual users have built products on Lovable and Replit that attracted users without venture funding, formal launches or dedicated product teams.
PYMNTS Intelligence found that 48% of online shoppers in surveys across the United States, Brazil and the United Arab Emirates used AI during their most recent purchase. The same consumer adoption is now appearing on the supply side, as people use AI not only to compare products but also to produce software they could not find in the market.
This story draws on original reporting from PYMNTS.
