Data for Optimised Reasoning and Unified Modelling

Seque

Responsible anonymisation for AI-ready text

Seque helps teams prepare sensitive records, correspondence, field notes, and case data for large language model workflows without exposing personal or operationally sensitive information.

See how it works
Explore features
Back to products

How Seque works

1Where Seque runs

Every check and anonymisation happens in your local context. Only validated prompts are sent to AI applications, and Seque does not store your prompts.

Your computer

Your browser

Seque extension

Checks what you type, paste or attach on AI sites, in Chrome, Edge and other Chromium browsers.

Computer/Server

Seque agent

Seque agent extends your organisation’s AI policies beyond the browser, in sync with your extension setup, for users of desktop AI applications.

To the AI: only the checked message

AI platforms

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

and more, plus any internal AI tool your administrator adds.

To the audit log: what kind, when, where

Audit log

Records what kind of information was found, when, and on which site. Never the value itself.

Some desktop AI apps send their traffic in ways the agent cannot read yet, so treat those as unchecked. Product names and logos belong to their owners. Seque is independent and not endorsed by them.

2What happens to a message

Pick an example to watch it go through.

What you wrote

Summarise the main risks in this project plan and suggest three ways to reduce them.

What the AI receives

Summarise the main risks in this project plan and suggest three ways to reduce them.

Sent as writtenNothing sensitive was found, so the message goes through untouched. Most prompts look like this one, and Seque stays out of the way.

What you wrote

Draft a polite reply to thandi@acme.co.za confirming that the claim for ID 8001015009087 has been approved.

What the AI receives

Draft a polite reply to [EMAIL_1] confirming that the claim for ID [ID_NUMBER_1] has been approved.

Tab closed: cleared

Held in this tab’s memory only

[EMAIL_1] → thandi@acme.co.za

[ID_NUMBER_1] → 8001015009087

Cleared when you close or reload the tab.

  • Neversent to the AI platform
  • Neversaved on your computer
  • Neversent to Seque
  • Onlythe audit log notes that an email address and an ID number were masked, not what they were

MaskedSeque offers to swap the email address and ID number for placeholders, then send. The AI can still write the reply, but the real values never leave your computer and are never saved.

What you wrote

Book a return flight to Nairobi for our consultant, passport number A12345678, leaving on Monday.

What the AI receives

Nothing yet. Seque waits for you to edit the message, or to choose Send anyway if your organisation allows it.

WarnedSome information is sensitive but sometimes genuinely needed. Seque tells you what it found and lets you decide, so nothing goes out by accident.

What you wrote

Why does this keep returning access denied? My key is AKIAIOSFODNN7EXAMPLE

What the AI receives

Nothing. The message was not sent.

BlockedPasswords, API keys and private keys have no safe way to be shared, so the message cannot be sent. Seque explains why, and you can remove the key and try again.

Checked before it leaves

Typing, pasting and attaching a file are all checked on your computer, before anything reaches the AI.

No copy of masked values

The link between a placeholder and the real value lives only in your open tab. Nothing writes it to disk or sends it anywhere.

Clear about its limits

Seque does not read text inside images, screenshots or audio. Treat those as unchecked.

From sensitive source text to safer AI use

Seque structures anonymisation as a repeatable workflow so teams can reduce risk while preserving the context needed for analysis, summarisation, and knowledge work.

1

Ingest content

Bring in text from records, notes, forms, reports, and correspondence.

2

Detect sensitive data

Identify names, contacts, identifiers, locations, and contextual risk markers.

3

Replace safely

Apply consistent substitutions that protect identity while retaining meaning.

4

Review outputs

Support human checks where sensitivity, ambiguity, or impact is high.

5

Prepare for LLMs

Format anonymised text for prompts, summaries, analysis, and model workflows.

6

Keep an audit trail

Maintain transparency around what was changed, reviewed, and approved.

Designed for practical AI safety

Seque focuses on the operational controls teams need before sending real-world text into AI systems.

Structured anonymisation

Replace sensitive values consistently so outputs remain useful for analysis without revealing identities.

  • Personal details
  • Contact information
  • Identifiers and references

Context preservation

Retain the operational meaning of the text so AI workflows can still summarise, classify, and reason effectively.

  • Role-based replacements
  • Consistent placeholders
  • Readable anonymised output

Where Seque fits

Use Seque wherever staff use public AI tools with information that belongs to clients, employees or patients, from corporate firms to public-sector teams.

Law firms

Keep client contact details, ID and passport numbers, and company registration numbers out of public AI tools while staff draft, summarise and research.

Human resources

Protect employee ID numbers, tax references, bank details and home addresses when HR teams use AI for letters, policies and job descriptions.

Finance

Catch card numbers, bank accounts, IBAN and SWIFT codes, and VAT numbers before they reach a chatbot. Passwords and API keys are blocked outright.

Health

Catch medical aid numbers, patient ID numbers and contact details before they are pasted into an AI tool. Seque recognises identifiers, not clinical wording, so a diagnosis written in words is not flagged.

Citizen and community records

Prepare case notes, engagement records, and service delivery data for safer summarisation and analysis.

Research and field notes

Anonymise qualitative notes while preserving the structure needed for coding, synthesis, and reporting.

Operational correspondence

Reduce exposure in emails, reports, and internal text before AI-assisted drafting or analysis.

AI governance workflows

Create a controlled step between sensitive source material and large language model usage.

Prepare sensitive data with care

Seque supports responsible AI adoption by helping teams reduce privacy risk before text enters model workflows.


View pricing