Fixing the air we breathe: how one founder built a new indoor air system using OpenAI tools

Walk into any office, hospital, or home in an Indian city and the air conditioning is running. It keeps the room cool. What it does not do is make the air safe to breathe. Air conditioners recirculate the same indoor air, suppress oxygen levels, and do nothing about particulate matter, humidity, or ventilation. Purifiers filter but do not ventilate. Dehumidifiers address one variable while ignoring the rest. Every device solves a piece of the problem and creates a different one.

India’s air quality crisis has pushed this contradiction into sharp focus. Delhi, Gurugram, Mumbai. Cities where AQI regularly crosses 300, where parents keep children indoors during winter months, where hospitals see seasonal spikes in respiratory illness. Yet the devices people rely on for “clean air” were never designed to deliver it.

This is the gap that Abhinav Gupta has spent the last decade studying.

For 11 years, Gupta focused almost entirely on measuring air quality. His first company deployed thousands of monitoring devices worldwide and sold more than 5,000 units. . The work was serious enough that he received a funding of a million dollars to deploy thousands more sensors worldwide and study how air affects human health.

By 2024, he had exited the business.

What followed was not an immediate pivot but a return to the same unresolved question. Measurement had scale, data, and visibility. It did not fix the air people breathed.

“ACs cool air but degrade oxygen quality,” Gupta says. “People in air-conditioned environments are not necessarily breathing better. Often it is worse.”

That realization became Breethr. A single system designed to manage temperature, humidity, filtration, and fresh-air ventilation together rather than as disconnected appliances. The objective was simple in principle and complex in execution. Improve the quality of every breath indoors.

Engineering against contradiction

Combining ventilation, filtration, cooling, and moisture control inside one system creates immediate technical conflict. Cooling systems are designed to seal environments for efficiency. Ventilation requires bringing outdoor air inside. Each function works against the other in terms of energy and airflow.

Breethr draws outside air through a four-stage filtration process, regulates temperature and moisture, and introduces the treated air indoors with positive pressure. Even when doors or windows open, polluted outdoor air is pushed away rather than pulled inside. Gupta says the system can reach single-digit AQI levels in under twenty minutes.

Designing something that could do this reliably required far more than incremental hardware tweaks. Gupta turned to OpenAI’s tools as a working research partner during the engineering phase.

Research, iteration, and OpenAI

Gupta describes months of back-and-forth exploration using GPT-driven research and image generation to test concepts quickly. Instead of manually reviewing scattered academic literature, he could analyze multiple research papers, generate design hypotheses, refine mockups, and iterate in rapid cycles.

One unexpected example came from outside air systems altogether. While studying thermal control, Gupta asked how Formula One cars manage brake temperatures during races. The materials and techniques surfaced through that exploration eventually informed Breethr’s own engineering decisions.

“My role is research and growth,” he explains. “I use deep research for engineering questions, and the same system helps with market intelligence or operational planning. Work that once required a full team can now move much faster.”

OpenAI tools also reduced reliance on large software teams during prototyping. Gupta used them to build early firmware logic, application layers, and internal workflows independently before formal engineering scale-up.

From device to data layer

Breethr is not only a hardware system. Each installed unit captures continuous air-quality information. With thousands of planned deployments, Gupta sees the emergence of a real-time environmental dataset tied directly to human health outcomes.

That dataset opens a different line of thinking. Insurance underwriting based on actively managed air environments rather than passive exposure. Integration of pollen forecasts, wildfire trends, and public health signals. Modeling respiratory risk with far greater precision than traditional actuarial methods.

“I spent years studying air-quality data,” Gupta says. “Data alone does not change anything. You need systems that act on it.”

Early adoption and expansion

Breethr has already been deployed across more than one million square feet in India, including hospitals, IVF clinics, restaurants, commercial buildings, and residential homes designed for children and elderly occupants. Revenue has crossed $400,000.

In September 2025, Gupta spent time in the United States through a fellowship program, where he began testing the system with a housing developer exploring integration into new residential construction. At the same time, conversations around distribution have been emerging across markets such as Vietnam, Egypt, and the United Arab Emirates, even as the core deployment work continues in India.

“In India, this is about survival,” he says. “In the US, it is framed as quality of life. But unhealthy air is a global issue.”

The long-term ambition is straightforward. Make managed indoor air as standard as cooling itself. Homes that maintain stable internal conditions regardless of pollution, allergens, wildfire smoke, or extreme weather outside.

Acceleration through OpenAI

Across Breethr’s development, OpenAI’s role has been less visible than foundational. It shortened research cycles, enabled independent prototyping, and allowed a small team to explore engineering and data questions at unusual speed.

For Gupta, the shift is personal as much as technical. Eleven years of tracking air quality produced insight. Building Breethr required turning that insight into something tangible.

“OpenAI helped me move faster from understanding the problem to actually solving it,” he says.

Breathing has always been automatic. Ensuring the air behind it is safe has never been. Breethr sits inside that unfinished problem, attempting to close the distance between knowing and fixing.