PrivacyPal Cloud

Install once. Every AI workload on the box runs through PrivacyPal.

A privacy gateway you install on any desktop or server inside your network. Every agent, copilot, and AI tool running on that machine is automatically protected — no SDK changes, no model swaps, no third-party DPAs. PrivacyPal Cloud lives entirely in your private network.

# Install on any desktop or server in your network
$ curl -sSL https://get.privacypal.ai | sh

# That's it. Every AI agent running on this machine
# is now protected by PrivacyPal Cloud.

$ privacypal status
✓ Gateway running on 127.0.0.1:7878
✓ Watching 4 local AI processes
✓ Routing through privacypal.yourco.internal
✓ 0 prompts left this network
Built for your infra

Deploy anywhere your workloads already live.

Zero code changes

PrivacyPal installs on the host and intercepts AI traffic at the network edge. Existing agents, copilots, and SDKs keep their config — every request is twin-swapped on the way out.

AWS · Azure · GCP

Terraform, Helm chart, or single-container deploy. Production-ready in an afternoon.

Air-gapped option

Run detection with on-prem models. No outbound traffic. Perfect for regulated workloads.

Any provider, any model

Route to OpenAI, Anthropic, Google, Mistral, Llama, your own fine-tune. Switch per request.

Streaming-safe

Detects and swaps mid-stream without breaking token-by-token response. Users never see the difference.

Horizontal scale

Stateless workers. Scale to 100k req/s. P99 latency under half a second.

Included with Cloud

Plus a full Data Security Posture Management stack — built in.

Cloud doesn't just protect AI traffic in motion. The same install gives you continuous discovery, classification, risk prioritization, and remediation across every data store that feeds your models — running entirely inside your network.

Agentless discovery

Connect a data store and it's in the scan queue. SQL Server, PostgreSQL, MongoDB, Supabase, Drive, OneDrive, Dropbox, object storage — read in place, no agents, no re-architecture, no performance hit.

Context-aware classification

The same intelligence that powers Privacy Twins reads your data the way a human would. PII, PHI, financials, proprietary schemas — surfaced across structured and unstructured stores. No regex to maintain, no rule sprawl.

Risk that's actually risk

Sensitivity tied to access patterns, exposure, and business purpose. A customer SSN in a public bucket is critical; a fixture in a dev database is not. Your team stops chasing noise and starts moving the needle.

One-click remediation

Revoke access, mask columns, kick off a workflow, or route findings to the data owner with full context. Discovery without remediation is a report nobody reads — Cloud closes the loop.

RAG, vector & agent memory

Scan the AI data layer the way you scan the database layer: vector stores, RAG pipelines, graph databases, agentic memory. Where your models read from — and where sensitive data quietly leaks in.

Sovereign by design

Scans run inside your network, against the same gateway already protecting your AI traffic. Findings, classifications, and remediation actions never leave your tenancy.

"We shipped an AI assistant on patient records in six weeks. Legal signed it in two days. That's PrivacyPal Cloud."
— CTO, healthcare SaaS · Series B

Your stack.
Your keys. Your data.

Ship AI features your security team reviews in days, not quarters.

Talk to Engineering View docs