Please call back on +49 30 23125 417
Write
You write your prompt as usual. AEGIS reads along in the input field, on the device itself, and does not transmit the text.
The data protection filter for external AI.
BRANE AEGIS protects sensitive data right inside ChatGPT, Claude, Gemini, Microsoft Copilot and Grok, before it reaches the AI model. Your teams keep using their AI tools; AEGIS spots sensitive information as they type and enforces the rules you define centrally.
What it looks like
Protection right inside the AI tool.
BRANE AEGIS works where your teams already work. No new AI portal, no extra process: AEGIS lays its control layer directly over the AI use you already have.
Good evening
How can I help you today?
Claude can make mistakes. Please double-check responses.
Recreated interfaces. ChatGPT, Claude and Gemini are trademarks of their respective owners.
Sequence
Protection starts as you type.
- Please call back on +49 30 23125 417, invoice copy to reklamation@example.com.
Detect
AEGIS spots sensitive information right in the input field, here an email address and a phone number. This happens before the content goes to the AI model.
- 2 sensitive itemsChange my prompt↵Please call back on +49 30 23125 417, invoice copy to reklamation@example.com.
Protect
AEGIS enforces your policy: a warning with the number of findings appears above the field, and a click or Enter replaces the values with placeholders.
- 2 replaced — safe to sendPlease call back on [PHONE_1], invoice copy to [EMAIL_1].
Keep working
You stay in your AI tool. The prompt now says [EMAIL_1] instead of the email address. The workflow stays, the sensitive data doesn't.
Centrally managed
Configure once, centrally. Enforce everywhere.
Different AI. One central policy.
ChatGPT today, Claude tomorrow, then Gemini, Copilot or a new service: the same company policy applies. You add further AI services through the administration.
- Card number, Check digit
- IBAN, Check digit
- Phone number
- Tax ID, Check digit
- Pension insurance no., Check digit
- API key
- Address
- Date of birth
- Licence plate
- IP address· once enabled
Data from everyday work
From email addresses to API keys. Each kind can be switched on centrally; for structured data AEGIS also checks the check digit.
AllowWarnBlock
chat.example.com is blocked by your policy
Open an approved tool →Decide what is allowed
Allow, warn or block, per AI application: your policy becomes a technical rule. Blocked sites show the way to an approved tool.
Semantic detection
More than patterns.
Not every secret has a format: project names, internal systems or unreleased products don't match any rule for an IBAN or an email address. BRANE AEGIS understands what the prompt says.
Summarize where Project Northern Light Project codename stands for the steering committee, including the tolerances from drawing ZN 4471/03-b Your own data type and the prices from the Q4 supplier list Your list.
3 sensitive itemsChange my prompt
Names, codenames, internal terms
A semantic model on the device recognizes them in German and English.
Your own data types in plain language
Your administrators describe new data types in their own words. No classification project, no training.
Your list of confidential terms
Your company provides a simple list of what is confidential, once. AEGIS recognizes every entry in the input field.
Privacy
What your company sees. And what it never does.
With a company account, AEGIS reports counts to the console, not content. Everyone in the workspace can read what is recorded.
Visible in the console
- Which kind of data was detected
- On which AI site
- How often and when
Never transmitted
- The prompt
- The detected values
- The page and its address
Administrators can switch off each of these, including attribution to individual people. The database drops switched-off details before they are stored.
BRANE AEGIS + BRANE AIF
Control external AI. Enable secure AI.
One set of company rules, two ways to use AI safely.
Your company policy and your list of confidential terms
Defined once, centrally. Applied everywhere.
In the browser
BRANE AEGIS
For the external AI tools your people already use
- Employees write in ChatGPT, Claude, Gemini, Copilot or Grok as usual.
- AEGIS spots personal data and company terms, such as codenames and internal systems, right on the device.
- Sensitive values are replaced with placeholders before anything is sent.
- The external AI gets a safe prompt. The employee keeps working.
Too sensitive for external AI?
Continue on the secure path
In your own environment
BRANE AIF
For AI work that needs a controlled path
- Employees and company applications send prompts and documents.
- Content is inspected inside your environment, including PDFs, Office files and spreadsheets.
- BRANE chooses the right path for every request.
Stays inside
A local model answers. Nothing leaves the company.
Uses cloud AI safely
Sensitive values are pseudonymized and restored in the answer.
Use AI. Protect data. Keep control.
No AI ban
Ban it or let it run unchecked?
AI has long been part of everyday work. Sensitive information can end up in external AI models by accident, exactly where the work happens: in the AI's input field.
Block AI
- People lose productive tools.
- New AI services appear anyway.
- Actual use becomes harder to control.
Allow AI
- People can work productively.
- Sensitive information can go to external AI providers.
BRANE AEGIS is the third option: allow AI and keep it under control. Not blocking AI, but controlling the risk.
See BRANE AEGIS with your own use cases.
We show you how AEGIS works in the AI applications you already use, and how you configure your AI policies centrally and enforce them technically.