Last updated November 18, 2026
Access is decided by the database
Every read and write a person makes is checked in Postgres with row-level security, so what they can see does not depend on a screen hiding it. You can grant access to a whole sheet, specific rows or specific columns, to people or groups, and write rules based on a cell’s value. An API key is different: a workspace admin creates it for one table, and it can read (and, unless read-only, write) every row and column of that table, so share it like a password. The activity log is written by the database itself; no one can edit it from the app.
Signing in
You can sign in with Google, a magic link, or an email and password. You can turn on two-step verification with an authenticator app in Settings. Relationsheet’s own staff console requires two-step verification on every session that uses it. We do not offer enterprise SSO or SAML yet.
Where your data lives
Your account and everything in your workspaces (tables, records and files) are stored by Supabase in the United States (AWS us-west-1). The website and the server code that answers your requests run on Vercel, in its Washington, D.C. region. Traffic is served over HTTPS. Keys to the services we use stay on our servers and never reach your browser. We do not use third-party analytics or error-tracking services; page views are counted by Relationsheet itself.
We do not offer hosting in the EU or any other region yet.
AI and your data
- AI features run only when you or someone in your workspace uses one. We keep an automated inventory of every place in the product that can send data to an AI provider, and our test suite fails if a new one appears without being listed.
- AI run by Relationsheet uses Anthropic and TypeSafe, and no other AI provider.
- Relationsheet does not use your content to train AI models. Anthropic and TypeSafe both state in their terms that they do not train their models on data sent through their APIs.
- Anthropic deletes API data within 30 days, except content flagged for a policy violation, which it may keep longer. TypeSafe’s published terms do not state a fixed retention period. We do not have a zero-data-retention agreement with either provider today.
- A workspace admin chooses how AI may be used in that workspace: all AI, only AI run by Relationsheet (no connected keys of your own, no fallback from one provider to another), or no AI at all. The database enforces the choice, not the screen.
- An admin can mark a column Never send to AI. No AI step, AI decision or AI extract in an automation then reads its values, and an automation run that has read one cannot copy it into a column that is not marked. People who can see the column can still copy it by hand.
- If you connect your own AI provider with your own key, that provider’s terms apply.
- AI decide is a preview not yet available to customers. It sends the values of the columns the step’s author ticks to TypeSafe. If TypeSafe is not sure or cannot answer, the values go to Anthropic only when the person who built the step chose that.
- Structure my data lets you choose how much is shared before anything is sent; by default it sends column headers, rounded statistics and made-up sample values, not your real values.
Full detail of what each AI feature sends is in our Privacy Policy.
Keeping and deleting data
We keep your data for as long as your account is active. You can export, correct or delete it, and close your account, at any time. Edit history (who changed which cell, and when) is kept for 180 days. AI decision history is deleted after 90 days. You can export any view to CSV, Excel or JSON.
Our database is backed up by Supabase once a day, and each backup is kept for seven days. Uploaded files are stored separately and are not part of database backups. We have not yet tested a full restore.
Health information
Relationsheet is not designed for protected health information (PHI). Do not store PHI in Relationsheet. We have no business associate agreement (BAA) with any of our providers, and our email provider cannot sign one.
What we do not have yet
- No SOC 2 report, ISO 27001 certificate or HIPAA attestation. We will not imply one with a badge.
- No independent penetration test yet.
- No data processing agreement for customers published yet.
- No hosting outside the United States.
- No customer-managed encryption keys and no single-tenant deployment.
- No published uptime commitment (SLA).
Service providers
The providers that process your data on our behalf, and what each one receives, are listed on our subprocessor page. When that list changes, we update the page and add a dated entry to its change log.
Reporting a security issue
If you think you have found a security problem, email Connect@SkillEra.IO. Please give us a reasonable time to fix it before telling anyone else. We will not take legal action against good-faith research that avoids other people’s data and does not disrupt the service. Our machine-readable contact file is at /.well-known/security.txt.