Redact sensitive information before sending it to AI

Velonos is a desktop app for macOS & Windows that replaces personal data and confidential details in your prompts and documents with placeholders, then restores them in the AI's response. All happens locally on the device.

Works with

  • ChatGPT
  • Claude
  • Gemini

From your prompt to your reply

4 sensitive details sent to AI4 details to AI 0 sensitive details sent to AI0 details to AI

On your device

Your prompt

Summarise the dispute with Northwind Trading Ltd. Martin Keller from 14 Linden Street, London says the invoice was paid to GB29 NWBK 6016 1331 9268 19.

Names, addresses and account numbers leave your device unchanged.

Sensitive information is identified and kept locally.

At the AI provider

What the AI sees

Summarise the dispute with [ORGANIZATION_1]. [PERSON_1] from [ADDRESS_1] says the invoice was paid to [IBAN_1].

The provider receives your clients' details and may keep them.

The provider receives only placeholders.

Back on your device

Your reply

Martin Keller claims Northwind Trading Ltd was paid to GB29 NWBK 6016 1331 9268 19. Ask for the bank confirmation before you reply.

The reply comes back, and the details stay out there.

Original details are restored.

Curious how it works, step by step?

  1. Write a prompt, paste a text, or attach a document, as you would today.

  2. Before anything is sent, Velonos detects names, addresses, organisations and financial details and replaces them with placeholders. This happens on your device.

  3. The AI provider receives the substance of the request, but nothing that identifies the people or organisations involved.

  4. Placeholders are applied consistently, so the model can reason about the content coherently. The quality of the answer is preserved.

  5. The original information is reinserted on your device. You receive a complete answer. The provider never saw the underlying details.

A privacy layer for AI that runs on your device

Velonos uses its own AI model, developed and trained by Software Mansion.

  1. Original details come back in the answer

    Placeholders stay consistent throughout a conversation, so the model can follow the context, and Velonos puts the original details back into the response on your device.

  2. Your team keeps its AI tools

    It works with ChatGPT, Claude, Gemini and many other AI tools, and everyone keeps working in the same apps as before.

  3. Detection runs on your device

    Many AI privacy tools route prompts through their own servers. Velonos runs its detection model locally, fully offline, so your data goes only to the AI provider you choose, already masked.

  4. It protects what matters most

    It masks prompts, images, PDFs, Word and PowerPoint files, and scans before sending.

Pricing

Choose the subscription plan tailored to your needs

Pricing is unavailable right now We could not load the current prices, so purchases are paused for the moment. Please try again in a while or contact us.
  • Individual

    Privacy for your everyday AI work.


    Your desktop app includes

    • App for macOS and Windows
    • Your data never leaves your computer
    • Redact prompts, supported PDFs and scans
    • Word, PowerPoint and images supported too
    • Restore details in supported AI replies
    • Personal rules and custom terms
    • App updates
  • Recommended

    Team

    Manage AI use across your team.


    Everything in Individual, plus

    • Admin panel with admin and member roles
    • Invite members; assign and reassign seats
    • Shared redaction policies and custom terms
    • Token-usage estimates by member and AI service
    • Exportable usage reports
  • Enterprise

    Contact us

    Fit Velonos to your organisation.


    Everything in Team, with a tailored scope

    • Custom cloud or on-premises deployment
    • Detection-model fine-tuning
    • SSO and automated provisioning
    • Audit exports and security integrations
    • Custom integrations and deployment support
    • Dedicated onboarding and agreed support SLAs

AI subscriptions and API usage are billed separately.

Set up your team

Pick how many seats you need. You can change the number later in the console.

Total

,

Frequently
Asked Questions

  • It's far safer when the AI receives placeholders instead of your clients' details. Velonos replaces names, account numbers and other identifying information on your computer before the request is sent. Law firms, accounting firms and other teams can then use AI with confidential documents without revealing who they concern.

  • AI data loss prevention (DLP) is a set of safeguards that keeps sensitive company and client data out of AI tools like ChatGPT, Claude or Gemini. Velonos handles it on each employee's computer: it finds personal and confidential details in prompts and files and replaces them with placeholders before the request is sent, so your team can keep using AI every day.

  • Your AI provider receives your prompts and files, with placeholders in place of the detected data. Detection runs on your computer, and that's also where the original data and the activity log stay.

  • Yes, occasionally. Like any detection model, Velonos can miss a detail or replace something that didn't need replacing, and scans or unusual formatting can affect results. Use the test preview to check what will be replaced, add your own terms and exceptions, and review important documents before sending them.

  • With Enterprise, Software Mansion fine-tunes the detection model on sample documents from your organization, so it better recognizes the terms and data that matter to you, and helps you deploy Velonos across your team.

  • Those plans govern what the provider does with your data, but the provider still receives it. Velonos makes sure identifying details never leave the office.

  • Velonos sends redacted copies of PDFs, scans, Word and PowerPoint files, and images. If a file can't be processed safely, it's blocked rather than sent as it is.

  • Velonos currently supports Polish and English. We're actively working on adding more languages.

  • No, not by default. For Team licenses, we can set up usage statistics at your request, such as the number of tokens used per seat, and display them in your admin panel.

  • Velonos is developed by Software Mansion, which is certified under ISO/IEC 27001, the international standard for information security management. It's also built to help your organisation meet its own data-protection obligations, especially under GDPR.

  • Velonos works on computers with macOS and Windows operating systems.

Contact sales

Tell us about your team and how you work with AI. We will get back to you shortly.

We use your details only to reply to you.

This form is protected by Cloudflare Turnstile.