Sanitext
PII Detection & Redaction API

Detect and redact PII
in any text.

One API call finds names, emails, phones, addresses and secrets in 30+ languages — and hands back clean, masked text. We run the model ourselves, so there's no per-token AI tax, pricing is flat, and your data never touches a third-party LLM.

No credit card · 300K characters free

input.txt → /v1/redact

Patient Maria García, email maria@clinic.es, called from +34 612 345 678 about invoice ES91 2100 0418.


$ curl sanitext.app/v1/redact \
  -H "Authorization: Bearer pf_live_…" \
  -d '{"text":"..."}'

{ "redacted": "Patient [LASTNAME],
   email [EMAIL], called from
   [PHONE] about invoice [ACCOUNT].",
  "count": 4, "ms": 47 }

Own the model

Zero per-token cost

30+ languages

Not English-only

EU data residency

Nothing sent to 3rd-party AI

No data retention

We log counts, not text

How it works

PII detection and redaction in two endpoints.

POST /v1/detect

Returns every PII span with label, byte offsets and a confidence score.

{ "entities": [
    { "label": "EMAIL",
      "start": 20, "end": 36,
      "score": 0.99 } ],
  "count": 1, "ms": 31 }

POST /v1/redact

Returns your text with every span masked as LABEL — ready to log, store or send to an LLM.

{ "redacted":
    "Contact [FIRSTNAME] at
     [EMAIL].",
  "count": 2, "ms": 28 }

PII redaction software, minus the enterprise tax.

The market splits into free-but-self-host Presidio and quote-only enterprise. Sanitext owns the transparent, self-serve middle — published prices, an API key in 60 seconds.

Sanitext AWS Comprehend Azure PII Presidio (OSS)
Self-serve in 60s self-host
Transparent flat price metered metered free
One-call redaction offsets only DIY
Multilingual (30+) EN + ES EN-first
No per-token AI cost
EU data residency region region self-host
Pricing

Free to start. $29/mo when you grow.

300K characters free to start — no clock, no card. Paid plans from $29/mo — up to 2× cheaper than AWS or Azure, with no bill-shock.

Questions

What counts as PII? +

Sanitext detects personal data: first and last names, email addresses, phone numbers, postal addresses, dates, URLs, account and card numbers, and secrets like API keys and tokens — in 30+ languages. Each is returned as a labeled span with exact byte offsets.

Detection or redaction — which do I get? +

Both. POST /v1/detect returns the spans; POST /v1/redact hands back your text with every span masked as [LABEL]. Same call, your choice.

Is my text stored? +

No. We log request metadata only — character counts, entity counts and timing — never your text or the PII itself.

How fast is it? +

Short inputs (1–3 pages) return in well under a second. Large documents take longer; the Scale and Enterprise tiers add an async bulk endpoint for them.

How is this cheaper than AWS or Azure? +

We run the open-weights model on our own infrastructure, so we pay no per-token AI fees — and we pass that on. Flat tiers, overage up to 2× cheaper, and a permanent free tier.

Is this GDPR or HIPAA compliance? +

It is a redaction and data-minimization aid, not an anonymization or compliance guarantee. Use it as part of a broader privacy-by-design approach. EU residency and a signed DPA are available on Enterprise.

Redact your first document in 60 seconds.

Sign up, copy your key, curl the API. No sales call, no cloud account.

Get your free API key