Swiss-made software

Public AI chats have your data.
Your employees are providing it.

Protect sensitive data inside ChatGPT, Claude, Gemini, Copilot and more, the moment your team starts typing.

Eclipsery detects sensitive entities in your prompt and swaps them for placeholders. You get the answer; the provider gets nothing personal.

Draft a reply to Marco Rossi at marco.rossi@finlombarda.it about the invoice due on 14 March, and confirm the transfer to CH93 0076 2011 6238 5295 7.

{n} entities masked before sending

The placeholder-to-original mapping stays in your browser for the duration of the request, so the answer you read is de-anonymized locally.

The risk

Do you know what your team shared with ChatGPT this week?

Employees share contracts, client records, and internal documents with AI tools every day, often through personal accounts nobody at the company can see. Once that data leaves your organization, there is no getting it back.

89%

of employees in Switzerland already use AI tools in their daily work.

66%

of Swiss companies have no policy for how AI can be used.

Every 3 days

employees share sensitive company data with ChatGPT, Claude or Perplexity.

Sources: EY Switzerland (2026), AXA Swiss SME survey via smerevision.ch (2025), Cyberhaven (2026).

How it works

Stop sensitive data from leaving your organization with eclipsery.

From a single sentence to entire documents: anonymize what's sensitive before it ever reaches ChatGPT, Claude, Gemini and more.

Browser extension

Anonymizes directly inside Claude, ChatGPT and any website, right from your browser.

Text anonymization

Masks names, emails, phone numbers, addresses, dates of birth, IBANs and other sensitive data before they reach an AI.

Document anonymization

Upload a document and get it back with every sensitive entity redacted, structured and unstructured data alike.

  • DOCX
  • PDF
  • XLSX
  • PPTX
  • TXT
  • MD

Privacy dashboard

See what's being protected, entity types and volumes over time, never the content itself.

Works with the AI tools your team already uses

  • ChatGPT
  • Claude
  • Gemini
  • Copilot
  • Perplexity
  • Grok
  • DeepSeek

Why eclipsery

See what eclipsery is catching.

Every protected sensitive item gets counted, but the content never gets read or stored.

Global (all companies)

Anonymized entities

13,230last 30 days

Text

3,57427%

Documents

9,65673%

Over time

Entities anonymized per day, peaking at 1,842 2,000 1,842

2026-07-122026-07-31

By entity type

  • Text
  • Documents
Organization 3,691
Product 2,428
Person 1,883
Sensitive date 1,046
Entities anonymized by type, last 30 days
By entity typeTextDocumentsTotal
Organization6133,0783,691
Product5771,8512,428
Person5171,3661,883
Sensitive date6224241,046

Sample figures. Your dashboard shows your own organization's numbers.

A few of the industries we protect

Not your industry? Eclipsery works wherever sensitive data meets AI.

Public administration

Constituent records, case files, and internal correspondence, protected the moment your teams use ChatGPT, Claude, Perplexity & Co.

Healthcare & pharma

Patient records, clinical data, and research findings, kept confidential across every AI interaction.

Financial services & legal

Client data, account details, and case files, protected against exposure to third-party AI models.

About us

Six years. 40+ AI projects. 30 clients.

Since 2020, Artificialy has built AI systems for organizations in various sectors, for example in banking, logistics, energy and utilities, and enterprise software. Eclipsery is the product of that experience: teams wanting to use public AI chats, but worried about their data. We built the solution to that problem.

FAQ

Frequently asked questions

It depends on what you enter. The revFADP applies directly to AI tools, and entering identifiable personal data into a public AI model can count as an uncontrolled data transfer you're responsible for. Anonymizing that data before it leaves your browser is what keeps everyday AI use within the law.

Names, personal identifiers, ID numbers, nationality; emails, phone numbers, addresses, postal codes; locations, physical addresses, geographic references; medical data, sensitive dates, personal deadlines; organizations, company identifiers, job roles; amounts, banking identifiers, payment references, transactions; passwords, security codes, token keys; various identifiers (invoices, contracts, orders, etc.), license plates, operational references, and many others.

Eclipsery is Swiss-built and Swiss-hosted, with maximum protection at the foundation. Most tools only catch what you type once you hit send, but by then the leak has often already happened: many AI platforms send partial data as you're still typing, through autocomplete, drafts, or background requests. Eclipsery blocks these partial calls too. It can be rolled out company-wide by IT through SSO with no manual installation per user, and allows a margin of personalization, for example excluding specific terms from masking. Nothing is stored: not even we, as the provider, ever see the original values that were masked.

No. Masking happens in milliseconds as you type or upload, and the AI's reply is de-anonymized back to the original values locally, in your browser. The conversation reads exactly as it would without eclipsery.

Detection keeps improving, but no automated system is perfect. You can always review and edit the masked version before it is sent, and flag anything it should have caught. eclipsery doesn't aim to replace your own company's privacy policy or AI usage guidelines, where these already exist, but to add an extra layer of protection.

Eclipsery is built for individuals and teams alike. Get in touch and we'll help you find the right plan for how you work.

Detection runs through infrastructure operated by Artificialy, a Swiss company. Nothing you anonymize is stored. Only anonymized statistics are kept, and never the values themselves.

Contact

Stop sharing client data
with someone else's model.

Tell us how your team uses AI today. Here's what Artificialy gives you:

What to expect

  • A real person, every time. No chatbots.
  • A personalized demo, under NDA if you prefer.
  • Your data stays in Switzerland, built and operated to Swiss data protection standards.

For information on how we process your data, please refer to our Privacy Policy.