Co-parenting data is among the most sensitive information that exists: children, legal agreements, financial records. We built just parent around that reality from day one.
We make money from subscriptions, not from selling your attention or your data to advertisers. This is a promise, not a phase - it's how the business is structured. If we can't survive on subscription revenue, we don't deserve to exist.
When you use Calm Assist or Draft Helper, your messages are sent to Anthropic's API - not stored, not used to train any model. We use Anthropic's data controls to enforce this. Your words stay yours, and they stay private.
Your family's data sits in a completely isolated environment. No AI feature, no engineer, no system can query across family boundaries without explicit authorization. It's enforced at the architecture level, not just policy.
All messages are timestamped when sent and cannot be altered afterward - not by you, not by your co-parent, not by just parent. The record is exactly what was written. That's what makes it court-admissible.
We don't sell, rent, or broker your data to any third party - including data brokers, advertisers, or analytics platforms.
We don't use your co-parenting messages, journal entries, or agreements to train AI models.
We don't run third-party advertising trackers. Zero. Not even 'anonymous' ones.
We don't build behavioral profiles to monetize. The only product we sell is the software.
We don't store AI conversations beyond the technical minimum to generate a response.
Each family's data lives in a completely separate tenant context. The database query layer enforces family-level boundaries on every request - there is no "super query" that can cross family lines without explicit authorization.
This means a bug in one family's session cannot leak data from another. AI features only have access to the data you explicitly include in a request - they cannot browse or retrieve other records.
Enforced at the architecture level means it's not a rule we follow - it's a constraint the code makes structurally impossible to violate.
No cross-family data access, by design
When you use Draft Helper or Calm Assist, you explicitly trigger them. They don't passively scan messages, monitor your journal, or run in the background. Nothing goes to an AI without you clicking a button.
No storage at Anthropic. No training on your content. The API call is transient.
Read our privacy policy in full. Ask us specific questions. We'll answer them.