Air quality compliance runs on documents. A single Title V permit can hold dozens of emission limits, monitoring requirements, recordkeeping obligations, and deadlines, spread across 50 or more pages of regulatory language. For decades the only way to turn that document into a working compliance program has been to read it, line by line, and rebuild it by hand in spreadsheets.

That process takes a senior reviewer 30 to 40 hours per complex permit. It gets repeated at every renewal, every modification, and every consultant transition. And the output, usually a spreadsheet, starts going stale the day it is finished.

Today we are launching AirComply to end that.

What AirComply does

AirComply is an AI platform purpose-built for one domain: air quality permitting and compliance. Nothing else. It does four things.

Deconstructs permits. Upload a permit and AirComply produces a complete condition register in about 3 minutes: every enforceable condition, emission limit, monitoring requirement, and deadline, organized by emission unit, each item cited back to the permit page and condition number it came from.

Tracks obligations. The register becomes a live compliance system: a deadline calendar, rolling emission totals monitored against limits, and alerts when a 12-month total drifts toward a cap, before the exceedance, not after.

Generates documents. Semi-annual monitoring reports, deviation reports, annual compliance certifications, and more than 20 other document types, pre-filled from tracked data and staged for the Responsible Official's signature.

Answers questions with citations. Ask what happens if you exceed a VOC cap and get an answer built from your own permit's conditions, with the citations to check it.

Why we built it this way

General-purpose AI chatbots are impressive and unreliable in exactly the wrong ways for this field. They recall regulations from training data that may be years old, they do arithmetic inside a language model, and they state conclusions without sources. That is disqualifying in a domain where a wrong citation becomes an enforcement problem.

AirComply pairs frontier AI models with prompt systems, retrieval, and guardrails engineered exclusively for air quality compliance, and it holds to four rules.

First, every number and requirement cites its source: the permit line, the CFR section, the calculation input. Second, regulations are retrieved live from official sources at the moment of the task, never recalled from model memory. Third, calculations run in code. The language model orchestrates; arithmetic executes deterministically. Fourth, nothing is submitted on AI authority. Every output package includes a verification checklist for a licensed professional to review and sign.

AI does the reading. People do the judging. That division of labor is the product.

Three workspaces

Compliance looks different depending on where you sit, so AirComply ships with three workspaces. Facility teams get a dashboard built around their permit: deadlines, emission watch, document generation. Consultants get a multi-client view: every deadline across every client, deliverable drafting, and analysis tools like top-down BACT support. Agency reviewers get a queue view: completeness checks, cross-references against RBLC precedent, and review clocks.

You can watch all three run on our homepage demo, or work the tools yourself in the live workspace.

What happens next

AirComply is available now through early access. We onboard each organization by deconstructing a real permit together, which doubles as the fastest product demo in the industry: bring your permit, and leave with your compliance program rebuilt.

Request access and see what your permit looks like after 3 minutes.