What we collect, and why.
Kith turns your own relationship history into something Claude can think with. To do that we process data about your network — so we're deliberate about what we hold, where it goes, and what we never do with it. This page explains it in the same plain language we'd use in person. Pair with the terms of service for the full picture.
What we store about you
When you sign up we store the basics: your email address, display name, and an avatar URL. Sign-in is handled by Clerk. If you connect a mailbox, the OAuth tokens are stored by Clerk on our behalf — we never hold them ourselves — and we use them to read your mail metadata and contacts (see below). Your preferences and rules (cadences, voice, segments) live in our own database.
What we read from your mailbox
If you connect Google or Microsoft, we request read-only scopes only (gmail.readonly + contacts, or the Microsoft Graph equivalents) and use them to populate your network:
- Message metadata, not bodies — we read headers only: who a message is from and to, the date, the subject line, and the direction. We do not read, store, or transmit the contents of your emails.
- Saved contacts — we read your saved contacts to seed people into your graph.
- Two-way only — we only add someone when they're a saved contact or you've actually corresponded with them, so newsletters and no-reply senders stay out.
You can disconnect a mailbox at any time from Settings (and revoke the grant in your Google/Microsoft account). Disconnecting stops all future syncing.
Your relationship graph
From the above (and anything you add by hand or via Claude) we build your private graph: people (name, role, company, tags, a relationship-strength score, last-touched date), their identities (email/phone), interaction touchpoints (a type, a one-line summary such as a subject line, and a date), market signals, inbound engagement, and voice samples of your own writing. It's yours; we store it so the product works and expose it to Claude when you ask.
Personal API tokens
If you create a Claude / API token from Settings → Claude we store a salted SHA-256 hash of the token, never the raw value. We can't recover a token after it's shown to you — if you lose it, generate a new one. Tokens can be revoked at any time.
AI processing — what goes to Claude
Kith uses Anthropic Claude (via the official Anthropic API) to: parse free-text you dictate into structured contacts and rules; answer “who do I know who…” recall queries; assemble pre-meeting briefs; and draft notes and market-view pieces in your voice.
What we send for a given call is scoped to that task — the text you typed, the relevant slice of your graph, and (for drafting) your voice samples. The mailbox sync itself sends nothing to Claude — it's headers-only and never leaves our processing. Anthropic processes the request and doesn't train on it per their commercial terms.
Email delivery — Resend
If we send you product email (e.g. account or digest mail) we use Resend. Resend sees the recipient address and the message we send. We don't use it to email the people in your network on your behalf.
Analytics — Vercel
We use Vercel Analytics for pageview counts and a few product events. No cookies, no cross-site tracking — Vercel uses a privacy-friendly daily-rotating visitor hash. You can opt out per their privacy policy.
Who we share data with
Your data is processed by the following third-party services, each scoped to a specific job:
- Clerk — authentication, session management, and OAuth token storage. Sees your email, name, avatar, and connected mailbox tokens.
- Anthropic (Claude API) — AI processing. Receives only the slice of your graph and text relevant to the current task. Does not train on your data per their commercial terms.
- Resend — transactional email delivery. Sees the recipient address and message content for product emails we send you.
- Vercel — hosting and privacy-friendly analytics. Serves the app and collects anonymised pageview/event data with no cross-site tracking.
- Neon — managed Postgres database. Stores your graph, preferences, and account data, encrypted at rest.
- Cloudflare R2 — object storage for scanned business cards and screenshots. Files are encrypted at rest and served only to you via short-lived signed URLs.
We do not sell, rent, or otherwise disclose user data to any party not listed above.
Where it's hosted
The app runs on Vercel and your graph lives in a Neon Postgres database, scoped per user. Business cards or screenshots you scan are stored in Cloudflare R2 — a private bucket, encrypted at rest, served only to you over short-lived signed links. Connected mailbox tokens are held by Clerk, not us.
How we protect your data
- Encryption in transit — all connections use TLS 1.2 or higher. No data travels over the wire unencrypted.
- Encryption at rest — your database (Neon) and object storage (Cloudflare R2) both use AES-256 encryption at rest.
- OAuth token security — tokens for connected mailboxes are stored and managed by Clerk, never in our own database. We don't have direct access to the raw token values.
- API token hashing — personal API tokens are stored as salted SHA-256 hashes. We can't recover a token after it's shown to you.
- Voice sample security — voice samples of your writing are stored in the same encrypted database as the rest of your graph data and are only surfaced to Claude when a drafting task requires your voice.
- Access controls — every request is tied to an authenticated session. Your graph is scoped to your account; no other user can access it.
How long we keep things
- Profile, preferences, graph — until you delete your account.
- Mailbox sync state — a cursor + status per connected source; cleared when you disconnect.
- Tokens — until you revoke them, with a “last used” timestamp so you can spot stale ones.
- AI prompts + responses — sent to Anthropic per-request; we don't log the request bodies on our side.
Your rights
- Disconnect a mailbox at any time from Settings.
- Edit or delete any person or interaction in your network.
- Revoke any API token from Settings.
- Request export or deletion of all your data — email [email protected] and we'll come back within 30 days.
What we don't do
- Sell your data. Ever.
- Show ads.
- Train AI models on your data — neither us nor Anthropic (per their commercial API terms).
- Read or store the contents of your emails — we process headers only.
- Email the people in your network on your behalf without you doing it.
- Drop third-party tracking cookies or fingerprint across sites.
Children
Kith isn't designed for users under 16. We don't knowingly collect data from children. If you think we have, email [email protected] and we'll delete it.
Changes to this policy
We'll update this page whenever we change a third-party service or a data category. Material changes (new sub-processor, expanded data scope) will be flagged via email to your account address.
Contact
Privacy questions, data requests, or anything that smells off: [email protected]. We're a small team and read every message.
Last updated: 11 June 2026