Sitrستر

Use any AI model.
Expose nothing.

كل الذكاء الاصطناعي. لا شيء من بياناتك.

Sitr sits between your team and every LLM. PII and confidential figures are replaced with pseudonyms before the model sees them. The answer comes back complete, with real values restored.

sitr detect

Input

Approve contract for Ahmad Al-MansouriPerson, QID 28845001234QID. Send to hr@company.qaEmail.

What the LLM sees

Approve contract for [PERSON_1], QID [QA_QID_1]. Send to [EMAIL_1].

Person92%QID95%Email99%
Financial ModeNew

Your AI doesn't need your numbers.​The math is still exact.

Sitr hides revenue, salaries and deal sizes from the model, then computes the answer locally with your real figures. The model reasons about structure. Sitr does the arithmetic. Your numbers never leave.

You type

Revenue:

QAR 48.2M

Costs:

QAR 31.5M

What is the operating margin?

Model sees

Revenue:

[AMOUNT_1]

Costs:

[AMOUNT_2]

{{calc: ([AMOUNT_1]-[AMOUNT_2])/[AMOUNT_1] | percent}}

You get back

Revenue:

QAR 48.2M

Costs:

QAR 31.5M

Operating margin

34.6%

Computed by Sitr. Not guessed.

BanksInvestment firmsFinance teamsFund managers
Try the financial demo →

Every prompt is a data transfer.

When you send a message to OpenAI or Claude, that data leaves your network. It may never fully leave theirs.

They train on your prompts.

Without an enterprise contract, AI providers can use your conversations to improve their models. Your employees' QIDs, medical notes and financial details may end up permanently embedded in a system you do not control and cannot audit.

Your data crosses borders.

OpenAI, Anthropic and Google process data on US servers. Every prompt containing a Qatari QID or employee record is a potential PDPPL violation. Sitr removes sensitive values before the request leaves your network.

One endpoint. Full control.

Point your existing API calls at Sitr. It intercepts, strips PII, forwards to your chosen model, then restores real values in the reply. Your provider key. Your model. Zero new contracts.

How it works

1

Send

Your app sends a prompt or a block of text to Sitr.

2

Detect

Sitr finds every name, QID, phone number, IBAN, amount and address.

3

Mask

Each entity is replaced with a consistent placeholder. The LLM never sees real values.

4

Restore

Real values are put back into the reply before it reaches your app. Formulas are computed exactly.

All 6 GCC countries

Plus common entities: email, credit card, IP address, person name and passport. And now: monetary amounts, salaries, deal sizes.

QAQatar
QIDPhoneIBANPassportCRAddress
SASaudi Arabia
National IDIqamaIBANPhone
AEUAE
Emirates IDIBANPhone
BHBahrain
CPRIBANPhone
KWKuwait
Civil IDIBANPhone
OMOman
Civil IDIBANPhone

Simple pricing

Start free. Scale when you need more.

Free
$0/month
  • ·10,000 API calls/month
  • ·3 team members
  • ·Redaction API
  • ·All 6 GCC country packs
  • ·Community support
Start for free
StarterMost popular
$49/month
  • ·50,000 API calls/month
  • ·10 team members
  • ·LLM Gateway + Financial Mode
  • ·All 6 GCC country packs
  • ·Email support
Get started
Growth
$299/month
  • ·500,000 API calls/month
  • ·50 team members
  • ·LLM Gateway + Financial Mode
  • ·All 6 GCC country packs
  • ·Priority support
Get started

Ready to protect your prompts?

Free tier, no credit card required. API keys in under a minute.