Plain.
Try it
Plain · honest security

How we treat your data.

No platitudes. Three lists: what we do now / what's still coming / what we'll never do. Until we're large enough for full compliance audits, this page is the commitment.

What we do now

Today

  • Data lives in Cloudflare D1 and R2

    Documents and account data are in Cloudflare D1 (SQLite); images and uploaded assets are in Cloudflare R2. Both are encrypted at rest by Cloudflare (AES-256) and in transit over TLS. Export anytime to .pptx / .docx / .xlsx, or share as a link (the primary artifact).

  • Deleting means deleting

    Deleting an artifact removes the record outright — the link stops resolving and anyone holding it loses access. Account deletion is handled by request today, not yet a self-serve button: email hi@inplain.app and we erase your documents, assets, and chat history, then confirm when it is done. A self-serve delete is on the roadmap below.

  • How long we keep things

    Anonymous share links stop working 7 days after publishing — the page returns an expired notice and the content is no longer served. Signed-in artifacts are kept until you delete them or close the account. Billing records are retained as long as tax law requires. Prompt content is never stored at all — see below.

  • All traffic over TLS

    Web, CLI, and MCP all run over HTTPS. When LLM calls flow through our gateway we don't log prompt content — only usage metadata for billing (token counts, durations).

  • Magic-link sign-in, no passwords

    Better Auth + Resend. Sign-in links and one-time codes expire in 15 minutes and are single-use. OAuth via Google is also available. Sessions roll every 7 days.

  • Your content never trains a model

    We don't feed your documents to any LLM for training. Model providers run in zero-data-retention mode, so prompts and completions aren't stored on their side either.

  • Who else touches your data

    The complete subprocessor list: Cloudflare (hosting, D1, R2), DeepSeek (model inference, zero-data-retention), Resend and Brevo (transactional email), Creem (payments — card details go to the processor, never to us), PostHog (product analytics). Changes are published here before they take effect.

  • Gateway-only LLM access

    All AI calls go through Plain's gateway. No BYOK — it keeps the security model simple and auditable.

On the roadmap

Not yet, but coming

  • Self-serve account deletion

    Deletion works today but goes through hi@inplain.app and a human. A button in account settings that erases everything without asking us is on the roadmap.

  • SOC 2 Type II

    Audit materials in preparation. Targeting a Type II report before end of 2026. Customers who need SOC 2 to sign can email hi@inplain.app for an early vendor security questionnaire.

  • SSO / SAML

    Team plan ships SAML SSO + SCIM provisioning in 2026 Q4. Google Workspace OAuth works today as a stand-in.

  • Regional deployments

    Primary region is US-east today. EU / APAC regional storage is on the roadmap — ping us if you need it.

Never will

We will never

  • Sell your data

    Never. We earn from token billing, not data brokerage. Written into the terms.

  • Use your content for training

    Never. Even if an LLM vendor offers "opt-in training for a discount," we disqualify them in vendor selection.

  • Charge you to export

    .pptx / .docx / .xlsx / .pdf / Markdown source — always free. We sell AI compute, not data walls.

Found a security issue?

Email security@inplain.app. 24-hour reply, responsible disclosure 90 days. For non-urgent privacy concerns, hi@inplain.app.

Last updated · 2026-05 · this page tracks reality, not compliance theater