Why is AirNote Mac only?

AirNote is currently available only on Mac. If you use Windows, the obvious questions are: why, and when will that change?

The short answer is that AirNote’s privacy and price depend on running transcription on your own computer. Apple silicon gives us a fast, predictable platform for doing that well.

A short history of Apple silicon

Apple’s move towards designing its own chips began while Steve Jobs was leading the company. In 2010, the first iPad used the Apple-designed A4 chip. A decade later, Apple moved the Mac from Intel processors to Apple silicon, beginning with the M1.

That matters because Apple now controls both the hardware and software. M-series chips combine the main processor, graphics hardware, memory system and a Neural Engine—a part of the chip designed to accelerate machine-learning tasks.

The practical result is that a modern Mac can run focused AI models quickly and efficiently, without sending every task to a large data centre.

Why that matters for therapy transcription

An AI transcription model turns session audio into text. These speech-to-text models are smaller and more specialised than general-purpose AI systems: they have one focused job.

AirNote’s local AI transcription model runs directly on the therapist’s Mac. It uses Apple silicon to process long therapy sessions far faster than real time.

Most importantly, raw session audio is not uploaded to AirNote, Apple, OpenAI or another cloud provider for transcription. It stays on the Mac while the transcript is created and is deleted after successful transcription. For a conversation that may contain protected health information, or PHI, or special-category health data, that removes an unnecessary transfer of highly sensitive material.

AirNote does use secure cloud services for separate text-based AI features. Local transcription does not mean every AirNote feature is offline; it means the recording itself does not need to leave the device simply to become text.

Privacy and affordability are connected

A cloud transcription provider charges for processing every minute of audio. That cost repeats for every session and is ultimately reflected in the subscription price.

Local transcription uses computing power the therapist already owns. Once the model is included with AirNote, we do not need to buy server time for every minute of every recording. This helps us offer an ultra-affordable AI scribe for therapists without treating privacy as a premium upgrade.

Apple silicon therefore supports two important parts of AirNote at once: private therapy transcription and a sustainably low price.

Why not Windows?

Windows is not incapable of local AI. Newer AI PCs include neural processing units, or NPUs, which are specialised chips for AI work. The challenge is consistency.

Apple-silicon Macs share a common architecture and set of software tools. Windows laptops vary much more widely in processor, graphics hardware, memory, drivers and whether they have a dedicated AI accelerator. Some can run local transcription very well; many computers already in use cannot yet provide the speed and reliability we need for long therapy sessions.

Supporting Windows properly would require several technical paths, hardware checks and fallback rules. We could make it easier by sending recordings to the cloud, but that would add a raw-audio transfer and a per-minute cost. We do not want to compromise the reasons AirNote exists.

What about iPhone, iPad and Windows?

iPhone, iPad and Mac share the same broad family of Apple-designed chips and related software frameworks. That gives us a clearer path to bring AirNote to iPhone and iPad, and we are actively working on versions for those devices.

For Windows, we do not have a release date to announce. Our answer is not never. It is: when we can deliver the same standard of fast, accurate, local, private and affordable transcription on the Windows hardware we choose to support.

AirNote is Mac-only today because supporting a defined hardware platform lets us keep those promises now. It is an engineering decision made for therapists and their clients, not a lack of interest in Windows users.