Security you don't have to take on faith.
Proxyclinch encrypts your data in transit and at rest, keeps secrets in a KMS-backed vault, requires two-factor authentication for admin access, isolates every tenant at the database layer, and logs every sensitive action to a record nobody can edit. Here's exactly how.
Security enforced in the product, not left to a policy nobody reads.
Encrypted everywhere
Every connection to Proxyclinch runs over TLS, and your data is encrypted at rest. There's no plaintext path in between.
Secrets stay in a vault
Instagram channel access tokens and other credentials are stored KMS-encrypted, never as plain text in a database column — and they're revoked the moment you disconnect a channel.
Strong authentication
Passwords are hashed with argon2 and screened against known breaches. Two-factor authentication is mandatory for platform admin access, and available — mandatory for Owners on eligible plans — for your team.
Every tenant is isolated
Your data is scoped to your business at the database query layer, not just by convention — every tenant-scoped table ships with an automated isolation test in CI. Cross-tenant access is platform-admin-only and always logged.
An audit log nobody can edit
Role changes, member removal, channel connects/disconnects, publishes, exports, billing changes, and data-subject-request actions are all written to an append-only log — recorded, never altered or deleted.
We never see your customers' card numbers
Payments run through Paystack, which tokenizes and stores card data directly on its own systems. No PAN ever touches Proxyclinch's infrastructure.
One policy engine. Every outbound message, no exceptions.
Whether a message is sent by a person on your team, a keyword automation, or the AI assistant, it passes through the same single send path — there's no second, unguarded route out. That path enforces the 24-hour standard messaging window, per-channel pacing, and per-contact frequency caps required by Meta, and it always returns a clear reason when it blocks a send — never a silent drop.
- 24-hour messaging window
Enforced automatically on every send. - Per-channel & per-contact pacing
Token-bucket limits keep every account within Meta's rules. - Verified, de-duplicated webhooks
Every inbound event is signature-checked before it's processed once. - Truthful bot disclosure
“Are you a bot?” is answered honestly, always — no tenant setting turns this off.
Who else touches your data
Proxyclinch relies on a small, named set of sub-processors to run the product — never more than the request needs, and never your customers' card data.
| Sub-processor | Purpose | Data shared |
|---|---|---|
| Cloud infrastructure host | Runs the application, database, and encrypted backups. | All tenant and account data, encrypted at rest. |
| Meta / Instagram | The messaging channel itself — DMs, comments, and story replies. | Message content and contact identifiers for connected Instagram accounts. |
| Anthropic | Powers the AI sales assistant's replies. | Conversation content and your published knowledge base, sent only as needed to generate a reply. |
| Paystack | Processes subscription payments. | Billing details for your subscription. Card data is tokenized by Paystack directly — Proxyclinch never stores it. |
| Email / SMS provider | Delivers notifications, alerts, and one-time codes. | The contact address or phone number and content of the specific message sent. |
Security: frequently asked questions
Does Proxyclinch ever store my customers' card numbers?
No, never. Payments are processed by Paystack, which tokenizes and stores card data on its own systems — no PAN reaches Proxyclinch's infrastructure at any point.
Is my business's data isolated from other tenants on Proxyclinch?
Yes. Every tenant-scoped table is filtered by your tenant ID at the database query layer, not by a shared view or convention, and every one of those tables has an automated cross-tenant isolation test running in CI. Cross-tenant access is restricted to platform admins and is always written to the audit log.
Is two-factor authentication required?
It's mandatory for anyone with platform admin access, and available to every tenant team — mandatory for the account Owner on eligible plans, optional for other roles.
What happens to my conversation data when the AI generates a reply?
Only the conversation content and published knowledge-base material needed to generate that specific reply is sent to our AI provider, Anthropic. The AI never answers from anything else.
How do I report a security vulnerability?
Email security@proxyclinch.example with a description, reproduction steps, and any proof-of-concept — redacted of real tenant or customer data. We acknowledge reports within 3 business days.