Methodology
What is included and why
The guide starts with four setups: AirGradient ONE, Airthings for Business, Atmocube, Kaiterra Sensedge Mini. They were chosen because each publishes enough developer or support documentation to compare API access, data export and costs, and together they cover small projects and commercial deployments, local and cloud access. This is not a census of the market. Leaving a product out says nothing about its quality.
Adding a fifth setup, or substituting a product, requires a recorded reason. If a product is discontinued or can no longer be substantiated, its profile is withdrawn or clearly marked rather than presented as a current recommendation.
The comparison unit: device, platform and plan
We compare complete setups: a named device or model, the platform or account it reports to, and the plan that applies. A monitor and an enterprise platform are not interchangeable purchases, and one model’s sensors, gateway needs or API rights are never applied to the manufacturer’s other products.
Types of evidence
- Documentation review
- Reading the vendor’s product pages, developer documentation, support articles, pricing and terms. All current profiles rest on this alone.
- Vendor confirmation
- A written answer from the vendor to a specific question. None has been requested or received for the current profiles. If one is used, the profile will say so and date it.
- Hands-on testing
- Using the hardware or API ourselves. No hands-on tests have been performed. A page will describe any test explicitly if one is added.
Vendor claims are summarized in our own words and cited. They are not treated as independent validation of accuracy or performance.
How claims, conflicts and unknowns are recorded
Each setup is stored as a structured record. Every material claim has a status, a short summary, any conditions (plan, firmware, model or module), and at least one source with the date we checked it. The build fails if a claim lacks a source. The same records generate the comparison cards, the tables, the profiles and the downloadable register. Summary sentences and recommendations are written by the editor from those records and reviewed against them.
We use distinct states: available, conditional, not available, not publicly documented (we checked the vendor’s pages and it is not stated), unknown (we could not verify it, for example because a page did not load or sources conflict) and not applicable. A free account is never read as a free API, a local protocol is never described as a local REST API, and a public dataset is never presented as access to arbitrary buildings.
Where sources conflict, the profile states the discrepancy and what a buyer should confirm. Checked dates change only after the source is actually re-read. Product generation, firmware and region can matter, and are noted where the sources mention them.
Recommendations without scores
We do not use stars, numerical scores, “most popular” badges or an overall winner. Suggestions are conditional on a use case (small projects, commercial multi-site deployments, local access, cloud integrations) and explain the tradeoff. Where the evidence does not support a suggestion, we list the questions that would decide it.
Disclosures and paid relationships
Current relationships: The editor, Justin Carlson, co-founded and leads Invivus Technologies, which runs Moldmap, SensiReport and Moldlist and is developing Aerlume, an indoor mold sensor. He has no financial relationship with AirGradient, Airthings, Atmocube or Kaiterra, no vendor pays for coverage or placement, and the site has no affiliate links.
Paid relationships do not determine which setups are covered or the order they appear in; setups are listed alphabetically. If an affiliate or other paid relationship is added, it will be disclosed here and beside the affected link, and such links will be marked as sponsored. Only actual relationships are disclosed, not hypothetical ones.
Corrections and changes
Corrections can be sent through the channel listed on the About page. Please include the page, the statement and a link to the official source. Material corrections are applied to the underlying record and its checked date, and recorded below.
Change log
- November 7, 2026: initial publication of four documentation-based profiles.
Not a diagnosis, inspection or safety guarantee
Monitoring data is not a medical diagnosis, a mold inspection or a guarantee that a building is safe. Indoor Air API Guide is not an API provider, testing laboratory, certification body or medical service. For health concerns or building inspections, consult a qualified professional.