FAQ.

Afterthought is a private voice-first note capture app for Android. It turns fast voice dumps into editable local notes with transcript, title, summary, bullets, and tags.

The intended product architecture says no. Raw audio stays on the device and AI processing happens locally.

None. The app downloads a free open model to your device and runs commands against that model locally. That means Afterthought does not need ChatGPT or another hosted AI provider to process your notes.

No. Free open models that run on a phone are not as powerful as the largest cloud models. For basic chatting, note help, summaries, tags, and transcription support, they should still be fine for most people.

Afterthought is not selling metered cloud AI. The model runs on your phone, so the ongoing cost is hosting and delivering model files instead of paying a hosted AI provider every time you make a note.

You do. Your recordings, transcripts, summaries, tags, and notes are meant to live on your own device, not in an Afterthought cloud account.

There are no cloud subscriptions behind each note. The main ongoing cost is model hosting, which should stay negligible as long as new customers keep paying once for the app. I will keep updating models and shipping them for free for as long as suitable models are available. The goal is not to get rich from this; it is to help more people get useful AI on their phone without breaking the bank or selling out their privacy.

The app itself does not contain the model. It downloads the model separately, which means new models can be updated over time without requiring an app update. It also means bug fixes, improvements, and new features can be shipped without forcing every app update to be multiple gigabytes.