- The real data goes straight to the AI provider, unfiltered.
- No one in your company knows what sensitive information was shared, or with whom.
- Exposed to penalties under Chile's Law 21.719, with no way to demonstrate compliance.
The privacy layer for AI
Your team already uploads customer data to AI. Protect it before it leaks.
PrivIA detects names, RUT, emails, phone numbers and addresses in real time and replaces them with placeholders before they reach the AI. Your team works in ChatPrivIA with models from OpenAI, Anthropic or Google Gemini, and the sensitive data of your clients and your own employees stays protected, without the AI losing context.
Designed in Chile · Built for Law 21.719
Built for Law 21.719Sensitive data protected before it leavesReal-time tokenizationCorporate controlOpenAI, Anthropic and Gemini via API
The problem has changed
Blocking ChatGPT at the firewall no longer works.
“Shadow IT” has evolved into “Shadow AI” — and the risk is far greater. We used to worry about a file in a personal Dropbox. Today, people pay for their own AI subscriptions to work faster: they upload source code to debug it, paste customer spreadsheets without anonymizing them to summarize them, and if you cut off access at the office, they do it from their phone. All of it under IT's radar.
Security can no longer be “the department of NO” — banning by decree only pushes people to hide. The real challenge is to get ahead: provide a closed, secure corporate AI environment before the leak.
DETECTED ACTIVITYNOW
clientes_2026.xlsxUploaded to an external model, not anonymized
BLOCKEDContract summaryWith national ID, address and financial data
PROTECTEDPrivIA detects, tokenizes and returns the safe answer — without changing your team's workflow.
How PrivIA works
Your team uses AI. Sensitive data stays protected.
PrivIA protects the sensitive data of your clients and your own employees before it reaches the AI, without the AI losing context.
Employees
Trusted roles
Severance for Ana Rojas, RUT 12.345.678-5
writes
PrivIA Chat
Your team's secure chat
You · Severance for Ana…
AI · Done, Ana Rojas…
shows the real data on screen
real data
PrivIA Engine
Detects · Tokenizes · Encrypts · Restores
Ana Rojas → [PERSONA_1]
RUT 12.345.678-5 → [RUT_1]
encrypted in PrivIA
RUT 12.345.678-5 → [RUT_1]
encrypted in PrivIA
tokenized text
EXTERNAL AI PROVIDERSvia API
OOpenAI
GGemini
AAnthropic
Only get placeholders
The answer comes back restored, only on the employee's screen.
Names, RUT, emails, phone numbers and addresses are replaced before leaving. Every request is logged, so you can prove compliance.
5 types
of sensitive data protected
Context intact
the AI understands the whole conversation
Law 21.719
auditable record
Governance, not just access
Choose the AI provider. You keep the governance of your data.
Your organization chooses the AI provider and connects its own API key. PrivIA protects the sensitive data of your clients and your own employees before it leaves, and keeps a record so your company can prove it.
- OpenAI
- Anthropic
- Google Gemini
- via API, with your organization's key
- The AI provider gets a placeholder instead of the real data.
- Auditable record of what was protected, when and in which conversation.
- Your company can demonstrate governance under Chile's Law 21.719.
What does Law 21.719 say?
Chile's Law 21.719 applies from 1 December 2026. It classifies infringements as minor (warning or a fine of up to 5,000 UTM), serious (up to 10,000 UTM) and very serious (up to 20,000 UTM). For repeat offenses the fine can be tripled and, for companies that are not small businesses, reach 2% or 4% of annual revenue. Penalized companies stay on a public registry for five years. A bill under review in the Chilean Senate proposes postponing it to December 2027. This is general information, not legal advice: confirm the details with your legal team.
POST /v1/tokenize
{ "text": "Cliente Ana Rojas, RUT 12.345.678-5..." }
→ 200 OK
{ "text": "Cliente [PERSONA_1], RUT [RUT_1]..." } The same layer, in your own systems
PrivIA API: the data protection layer for any AI workflow.
ChatPrivIA is the ready-to-use corporate chat. The PrivIA API is the same detection and tokenization, integrated directly into your own systems — a CRM, an internal pipeline, your own agent — so any workflow that talks to an AI model first goes through the same protection layer.
Start risk-free
Choose how you want to be protected
ChatPrivIA
Private corporate chat, ready for your team today.
$19.990 CLP / month / user
reference price, subject to confirmation
- Corporate chat with real-time tokenization
- Models from OpenAI, Anthropic or Google Gemini, with your organization's API key
- Receipt or invoice, with an auditable record
PrivIA API
Embed the protection layer in your own systems.
Custom
based on volume and integration — still being defined
- Detection and tokenization API
- Integration with your own systems and workflows
- Implementation support
Your next prompt
Innovation shouldn't cost you your privacy.
Try ChatPrivIA for 15 days, no credit card, and connect your AI provider's API key in minutes.