Your data is yours. Here is exactly what that means.

What we store

Everything you brain dump, in your own words. What Corpus pulls out of those brain dumps: the people, projects and decisions you mentioned, and the connections between them. Your email address, and your first name if you gave one. That is the whole list.

Voice recordings and files you attach are transcribed or read, then deleted. Most recordings go straight from your browser to the transcription model. If the recording is long, or the first transcription fails, it is held in our storage for as long as it takes to transcribe, then deleted.

You can take all of it with you, at any time

Every capture and every page exports as plain Markdown. One file per topic, your words verbatim, the links between them intact. Markdown opens in Obsidian, in Notion, in any text editor, or in whichever model you prefer to think with.

Export is free, permanently, whether or not you ever pay us. A copy you can walk away with is the only version of “your data is yours” that means anything, so it is not behind a plan, a queue, or a support request.

You can delete all of it

Email privacy@mycorpus.net and we will delete your account and everything in it within 24 hours. No form, no exit survey, no offer to pause instead. If you want a copy first, export it or ask us and we will send one.

Who else touches your data

Corpus runs on other companies’ infrastructure. Here is where your content is hosted and why:

Supabase

Stores your database and handles sign-in.

Vercel

Runs the servers that answer your requests.

OpenAI

The model that reads your brain dumps to pull out people, projects and decisions, and that answers your questions. It also transcribes your voice recordings.

xAI

Replies in real time when you ask a voice question.

Resend

Sends reminder emails, which contain whatever you asked to be reminded about.

Inngest

Queues the background work that processes a capture after you save it.

Every one of them is contracted as a processor, which means they may use your data only to provide the service to us. None of them may train a model on it, sell it, or use it for any purpose of their own. We do not sell or share your data with anyone else, and there is no version of this business that depends on doing so. Charging for the product is what keeps that true.

Two things we will not overstate

A privacy policy that claims more than it can deliver is worth nothing, so here are the limits of the promise above.

OpenAI keeps API content for up to 30 days to monitor for abuse before deleting it. Within that window their staff can access it if they are investigating a specific abuse report. This is equivalent privacy to any company that has a contract with OpenAI that excludes model training, and many companies with sensitive data take this approach. Despite that, we aspire to eventually keep your data from ever even leaving your device, as model advances permit.

Access to production data on our side is limited to what is needed to run and debug the service. Nobody at Corpus browses your memories, and there is no internal tool built for reading them.

Legal process

Your text is processed on our servers, so a court order served on us reaches what we hold, and no wording on this page changes that. We will tell you if it happens, unless we are legally barred from telling you. This is true of every product that processes your data on a server, which is nearly all of them.

What we do not do

No advertising, ever. No analytics package, no tracking pixels, no third-party cookies. The only cookie Corpus sets is the one that keeps you signed in. We do not build a profile of you to sell, and we do not share anything with data brokers.

Changes to this policy

If we change anything here that materially affects you, we will email you before it takes effect rather than quietly updating the date at the top.

Contact

Anything at all: privacy@mycorpus.net.