One API for known AI models, with data cloaking built in.
Let your app use GPT, Claude, Gemini and other models through prepaid CloakGate credits. Before a request reaches the model, CloakGate can detect, redact, pseudonymize or block sensitive data.
No model provider account required.
Send a reply to Customer_1 at Email_1 about invoice Invoice_1. Card on file [BLOCKED_SECRET].
AI integrations leak more than you think.
Support tickets, CRM notes, user messages, invoices, logs and internal tools often contain personal data, secrets or business information. Without a gateway, those details can be sent directly to external AI models.
Cloak sensitive data before it reaches the model
CloakGate sits between your application and the AI model. It detects sensitive information and applies your rules before the request is sent.
Strip the sensitive value out entirely before the request is sent to the model.
Replace values with consistent placeholders, so the model still reasons about structure.
Stop the request entirely when it contains data your policy forbids from leaving.
Use models your developers already know
CloakGate does not hide AI behind vague tiers. Choose recognizable model families and see which model was used for each request.
General tasks, coding, text, reasoning
Long context, analysis, documents
Multimodal and Google ecosystem
Added by availability and region
No provider accounts. No scattered API keys.
Top up your CloakGate balance and use supported models through one API. Usage, policies and audit stay in one dashboard.
- ✓ One prepaid balance
- ✓ One API key for your app
- ✓ Spending limits per project
- ✓ Clear model costs, no surprise invoices
Audit AI usage without storing full prompts
CloakGate logs technical metadata: project, model, timestamp, token usage, detected data types and applied policy. Prompt and response content is not stored by default.
Integrate by changing your AI endpoint
Create a project, generate an API key, choose a model and set cloaking rules. Point your app at CloakGate and keep your existing request shape.
Final SDK and API details will follow the production implementation.
import { AIClient } from "cloakgate";
const client = new AIClient({
apiKey: process.env.CLOAKGATE_API_KEY,
baseURL: "https://api.cloakgate.com/v1",
});
// Point your existing AI calls at CloakGate.
// Sensitive data is cloaked before the model sees it.
const res = await client.chat({
model: "gpt", // known model families
policy: "strict-pii", // your cloaking rules
messages: [{ role: "user", content: prompt }],
});Where teams put CloakGate to work
Built for AI features that touch real customer data.
Hide names, emails, phone numbers and ticket references before sending messages to a model.
Summarize customer notes while replacing personal identifiers with placeholders.
Block secrets, API keys and internal credentials from ever reaching an AI model.
Process invoices and documents with numbers and parties pseudonymized.
Classify and route tickets without exposing raw customer content.
Draft sensitive messages while keeping identifiable data controlled.
Designed to reduce exposure, not create another data lake
We do not claim certifications we do not hold. CloakGate reduces unnecessary exposure and gives you control and audit — it is not a legal, security or compliance guarantee.
Simple pricing
Subscription plus prepaid AI credits. Pricing shown here is indicative during early access.
- ✓ 1–3 projects
- ✓ Basic cloaking
- ✓ Metadata audit
- ✓ Prepaid credits
- ✓ More projects & volume
- ✓ Advanced policies
- ✓ Team access
- ✓ Priority limits
- ✓ High limits
- ✓ More seats
- ✓ Audit export
- ✓ Custom rules
- ✓ BYOK
- ✓ Self-hosted later
- ✓ DPA
- ✓ Custom routing
Questions, answered honestly
No. The default CloakGate mode uses prepaid credits. You top up your CloakGate balance and use supported AI models through one API — no provider accounts on your side.
Yes. BYOK (bring your own key) mode is available on request for teams that want to use their own provider accounts. The default experience is prepaid CloakGate credits.
Prompt and response content is not stored by default. CloakGate can store technical metadata such as timestamp, project, model, token usage, detected sensitive data types and applied policy.
CloakGate is designed to detect categories such as names, emails, phone numbers, addresses, customer identifiers, invoices, API keys, tokens and other sensitive patterns. Detection depends on the policies you enable and language support.
No sensitive-data detection system is perfect. CloakGate is a risk-reduction and control layer. It helps reduce unnecessary exposure, apply policies and audit AI traffic — it should not be treated as a legal or security guarantee.
CloakGate adds a security check before the model call. In most AI workflows the largest part of latency still comes from the model response itself. Actual latency depends on prompt length, enabled policies and the selected model.
CloakGate is designed to support known model families such as GPT, Claude, Gemini and other providers depending on availability. The exact model list is shown in the dashboard.
Depending on your policy, CloakGate can redact the secret, replace it with a placeholder, or block the request entirely before it reaches the model.
Start using AI through a safer gateway
Prepaid credits, known models, sensitive data cloaked before it leaves your app. Join the developer preview.