Agencies
You handle your clients’ data: customer files, campaigns, contracts. A leak on your side is a leak on theirs, and your contract with them puts you on the hook.
Mission3-month trial across your whole fleet
AI reached the workplace faster than training did. Every day, employees paste data into it in good faith that should never leave the building. Tacite-IA stops it at the right moment, explains why, and teaches your teams the right habits. We are not after your wallet. We are after your reflexes.
Zero cloud Zero AI 3-month trial Built with GDPR in mind
ChatGPT Claude Microsoft Copilot Microsoft 365 Copilot Gemini Google AI Studio Le Chat Perplexity DeepSeek Grok Poe HuggingChat Claude Code Cursor Windsurf OpenAI Codex Gemini CLI Copilot CLI
Zero cloud.Zero AI.Zero surprises.
Zero cloud
Your messages and files are analysed on the computer. They never go to a cloud, neither ours nor anyone else’s.
Zero AI
No artificial intelligence reading your data: precise, verifiable rules and alerts everyone can understand.
Zero surprises
A 3-month trial across your whole fleet, nothing charged until it ends. After that, one simple price per user, with no hidden fees.
Only licences and updates go through our server, hosted in Europe: never the content of your messages.
Our mission
Tackling the misuse of AI. Not just blocking it.
Our goal is AI used responsibly: employees who know what they are sending, client, patient and employee data that stays protected, and organisations that can move forward with confidence. That is why Tacite-IA explains rather than punishes.
- Protect the people whose data is at stake, not just the companies.
- Teach at the right moment, rather than sanction after the fact.
- Build the tool with the people who use it, listening to those on the ground.
3 months to decide
- Month 1Pilot on a handful of devices
- Month 2Rollout across the fleet
- Month 3Review with the dashboard
Then you decide.
Card or SEPA Direct Debit required at sign-up, nothing charged during the trial, cancel at any time from your admin console.
What goes out never comes back
One copy and paste is all it takes. And the data has left the company.
A client’s name, an IBAN, a sick note: pasted into an AI assistant to “save time”, they end up on servers you do not control. Here is what happens, with and without Tacite-IA.
Message to the AI assistant
Write an email for Mr Johnson, GB82 WEST 1234 5698…, sick leave (depression)
AI assistant’s servers
Mr Johnson
GB82 WEST 1234 5698…
sick leave (depression)
- Stored on their servers
- Reviewed by people, depending on settings
- Reused to train models, depending on settings
Once a message is sent, it cannot be recalled. What happens to the data depends on each service and its settings, which your employees are not always aware of.
Files
A file goes out in full. Tacite-IA anonymises it first.
A customer export, a contract, a configuration file: dropped into an AI assistant, they take all their data with them in one go. Tacite-IA reads them on the computer, spots what is sensitive and sends only an anonymised version.
customer_export.csvtenancy_agreement.pdf.env
Spreadsheets (Excel, CSV), documents (Word, PDF, PowerPoint), text and configuration files: Tacite-IA reads their content on the computer, before they go out. A file that is unreadable or too large is flagged, never sent without warning.
Raising awareness
AI arrived without a user manual. Most teams have never been trained.
Few employees have ever been told what they must never share with an AI assistant. Tacite-IA does not punish: it explains, at the exact moment the mistake was about to happen. That is when a habit is formed.
- 01The reflexAn employee in a hurry pastes a case file into an AI assistant. Nothing warns them of the danger.
- 02The alert that explainsTacite-IA spots the data and explains why it is sensitive, at the exact moment the mistake was about to go out.
- 03The right reflexSame request, without the data. The AI’s answer is just as useful, and the lesson sticks for next time.
Executives, IT directors, DPOs, agencies
When data leaks, it is your organisation that answers for it.
Whether the data was sent by an employee, an intern or a contractor, from a laptop or over the company network, it is the organisation, as the party responsible for its clients’ and employees’ data, that is held to account, by the data protection authority and by its clients alike.
An employeein a hurry, acting in good faith, often never trained
A company computeror its network, or a work account
An AI assistantfree, consumer or business
And at the end of the chain: your organisation
- Controller of your clients’ and employees’ data (GDPR, Article 24)
- Required to protect it with appropriate measures (Article 32)
- Required to notify the supervisory authority of a breach within 72 hours (Article 33)
- Exposed to fines of up to €20 million or 4% of worldwide annual turnover (Article 83)
Executives
Banning AI rarely works: your teams are often using it already, sometimes with personal accounts. Governing how it is used is the most realistic path.
IT directors and DPOs
Tacite-IA gives you a concrete measure to point to: protection on every device, an enforced policy and an alert log, with no content ever transmitted.
The risk
Your teams already use AI. Your data goes with it.
Nobody pastes bank details into ChatGPT out of malice. All it takes is a habit, a deadline, one file too many.
One paste is enough
A customer spreadsheet, bank details, an HR report: a single paste into ChatGPT is enough for data to leave the organisation, with no way of getting it back.
Files too
A Word document marked “Restricted”, a salary spreadsheet, a PDF contract: dropped into a conversation, they are read in full by the remote service.
AI acts on its own
Coding assistants read files, run commands and send data. One .env file or one careless command, and the secret is gone.
How it works
Unobtrusive by design. There when it matters.
Tacite-IA only shows itself when it has to. The rest of the time, you will not even notice it.
You type, paste or attach
A message, a spreadsheet, a document: you work as usual, in the assistant of your choice.
Tacite-IA analyses it on your device
In a few milliseconds, before anything is sent. Numbers are validated against their check digits, so a plain order number does not trigger a false alarm.
You stay in control
Sensitive data? A clear window, masked values, and three choices.
AnonymiseSend anywayCancel
YouMessage, paste, file
Tacite-IAOn your device
AI assistantChatGPT, Claude, Copilot…
Live
A document is on its way to an AI. Here is what Tacite-IA sees.
Names, addresses, IBANs, health data, access keys: every item is spotted and replaced with a neutral tag in a few milliseconds, without sending anything to anyone.
Subject: Reminder: invoice F-2026-118
Client: Ms Jane Doe[NAME_1]
Address: 12 High Street, Manchester M1 1AA[ADDRESS_1]
Contact: j.doe@example.com[EMAIL_1]
Payment: GB82 WEST 1234 5698 7654 32[IBAN_1]
Note: on sick leave[HEALTH] until 12 October
API access: sk-live-9fT2…Qm4x[SECRET_1]
0sensitive item found
- NAME anonymised
- ADDRESS anonymised
- EMAIL anonymised
- IBAN anonymised
- HEALTH anonymised
- SECRET anonymised
Everything happens in the browser. The document never leaves the device.
0
data types monitored
identity, finance, health, secrets, markings
0
checksums verified
IBANs, cards, national identifiers
0
bytes sent to a server
analysis runs entirely on the device
0%
of websites covered
AI assistants blocked, other sites flagged
For businesses
Every profession has its own sensitive data. Tacite-IA knows them.
Eleven real-life situations, replayed just as your teams will experience them: HR, finance, legal, architecture, recruitment, accountancy, property… Click on a profession to see the scene.
Human resources
An HR officer is drafting a letter with Microsoft 365 Copilot.
Anonymised
The letter gets written, without the employee’s health or salary ever leaving the company.
Hello, how can I help you?
Press Enter to send
Take action
Your teams are using AI today. Protect them today.
Every day without a safeguard, customer, HR or financial data flows to public assistants. Tacite-IA installs in minutes and stays out of the way until the moment it matters.
For developers
One program. All your coding assistants.
Claude Code, Cursor, Windsurf, Codex, Gemini CLI, Copilot: assistants read your files, run commands and call tools. Tacite-IA checks every action on the device before it happens. With nothing extra to install.
Checked on the device, before sending
Codexcurl -F customer_export.csvBlocked
Claude CodeMessage containing an IBANAnonymised
CursorReading .envDenied
EdgeA client’s email addressFlagged
Protection
Far more than an email detector. A complete safeguard.
Eight families of protection, from card numbers to classified documents, all the way to the actions of coding assistants. In the main European languages, national identifiers included.
Personal and financial data
IBANs, bank cards, social security numbers, passports, email addresses, phone numbers, postal addresses.
GDPR Article 9
Health, religion, opinions, trade union membership, sexual orientation, ethnic origin, criminal convictions, minors: as soon as a person can be identified.
Classified documents
CONFIDENTIAL, RESTRICTED, classification levels and your own internal markings.
Technical secrets
API keys, tokens, passwords, private keys, connection strings.
Bulk leaks
A file or a response containing dozens of people is blocked, even with no other warning sign.
Word, Excel, PowerPoint, PDF
Document content is read on the device before sending: body, headers, comments, author.
Dangerous actions
For developers: mass deletion, unconditioned queries, files sent to external destinations.
Prompt injection
Instructions hidden in a web page or file to hijack the assistant: detected and flagged.
Wherever you work
In the browser. And in every coding tool.
Chrome and Edge extension
Messages, pasted content and files, checked before the website receives them.
ChatGPTClaudeCopilotMicrosoft 365 CopilotGeminiLe ChatPerplexityDeepSeek+ every other site
On AI assistants: a decision window before anything sensitive is sent.
On every other site: a discreet alert for serious data, without ever blocking.
Unapproved AI tools: a warning banner, or a block if your organisation decides so.
Coding assistants
For development teams who hand code and commands over to an assistant.
Claude CodeCursorWindsurfOpenAI CodexGemini CLICopilot CLIVS Code and JetBrains (Claude Code)
Messages and files read: sensitive data spotted, anonymised copy offered.
Commands: confirmation before a mass deletion, an outbound transfer or an unconditioned query.
Anti-bypass: the assistant cannot switch off the protection without your consent.
Prompt injection: booby-trapped content is flagged and the assistant is warned.
Pricing
3 months to try it. Then one simple price.
Test Tacite-IA across your whole fleet for 3 months: pilot, rollout, review. Nothing is charged before the trial ends, and you can stop at any time from your admin console.
Annual
€6excl. VAT per user per month
i.e. €7.20 incl. VAT
€72 excl. VAT (€86.40 incl. VAT) per user, billed once a year.
Try it for 3 monthsMonthly
€8excl. VAT per user per month
i.e. €9.60 incl. VAT
Billed monthly, no minimum term.
Try it for 3 monthsIncluded in both plans
- Browsers and coding assistants protected
- Analysis runs entirely on the device, no content transmitted
- Admin console, dashboard and reports
- Detection rules and software continuously updated
- Remote deployment (Intune, Group Policy, Jamf, Kandji)
- Email support
One user means one protected device (computer or session). Prices incl. VAT include French VAT at 20%. Minimum of 5 users. For more than 250 users, or for deployment support, get in touch and we will send you a quote.
How does the trial work?
The questions people ask us before signing up, with straight answers.
Why do you need a card at sign-up?
So that protection carries on without interruption if you adopt Tacite-IA when the trial ends. Nothing is charged during the 3 months. The card (or SEPA Direct Debit mandate) is stored by Stripe: we never see its number.
What happens after 3 months?
Your subscription starts, for the number of users you chose, which you can adjust beforehand. Without a valid payment method, the subscription ends and protection stops on the devices.
How do I cancel?
At any time, from the Subscription page of your admin console. If you cancel during the trial, nothing is charged. After that, cancellation takes effect at the end of the period already paid for.
Is my data used to pay for the service?
No. No advertising, no resale, no commercial exploitation. Analysed content never leaves your devices: we could not sell it even if we wanted to, because we never receive it.
What if we grow?
You can adjust the number of users whenever you like from your admin console, even during the trial (nothing is charged before it ends). After that, added users are billed pro rata. For more than 250 users, we will send you a quote.
Will I get an invoice for our accounts?
Yes. Every invoice, showing VAT, is available in your admin console. Your organisation’s VAT number can be included on it.
Remote deployment
Installed across the fleet. Without touching a single device.
Tacite-IA is deployed like the rest of your software, from your device management tool. Employees have nothing to do and can neither change nor remove it.
1
Your IT team deploys
With Intune, Group Policy, Jamf or Kandji: one package, your key, and you are off.
2
Every device is set up
The Tacite-IA program installs itself, enforces the extension and configures the coding assistants.
3
Everything is protected
Browsers and assistants are checked from the next start-up. Updates are automatic.
0 / 48 devices protected
By sector
Built for organisations that cannot afford mistakes.
Banking and insurance
Bank details, policy numbers, claim files: spotted before anything is sent. Advisers can still use AI to draft and summarise.
Health and social care
Diagnoses, treatments, national health numbers: health data stays put, even when it is buried in a report.
Public sector
“Restricted” documents, citizen case files, internal markings: one single policy enforced across the whole fleet.
Lawyers, notaries and consultants
Contracts, deeds and confidential case documents: blocked before they reach a public AI.
Architecture and engineering
Planning applications, client contact details and budgets, tender documents: specifications get written without exposing clients.
Accountancy
Ledgers, payslips, clients’ bank statements: no named account goes anywhere near an AI.
Property
Tenant files, ID documents, tax notices: affordability can be assessed without sending the supporting documents.
Recruitment and HR
CVs, dates of birth, salaries, appraisals: candidate summaries stay focused on skills.
Communications agencies
Customer files, confidential briefs, embargoed campaigns: AI helps you create without giving anything away.
Education and research
Pupil and student data, minors, unpublished research findings: protected from the moment they are typed.
Industry and R&D
Project numbers, drawings, manufacturing secrets: your own patterns are added to the standard detections.
Software vendors and IT services
API keys, .env files, test customer databases: teams code with Claude Code without exposing secrets.
Why Tacite-IA
Protection that requires no blind trust.
Protect your data by sending it somewhere else to be analysed? We chose the opposite approach.
| Criterion | Tacite-IA | Typical online solutions |
|---|---|---|
| Where is your data analysed? | On your device, and nowhere else | Often on the vendor’s servers |
| Data passed to a third party | None, zero telemetry | Content or extracts often sent for analysis |
| Set-up | A few minutes, via browser policy | Often an integration project |
| For the user | They choose: anonymise, send or cancel | Often a block, with no alternative offered |
| Coding assistants | Claude Code covered natively | Coverage varies |
| Designed for | GDPR and European confidentiality markings | Often generic frameworks |
IT directors, CISOs, DPOs
Designed to be deployed, not endured.
An audit-mode pilot, a policy that fits your organisation, one-click deployment across the whole fleet. And users who understand why.
Let teams make the most of AI, without ever losing control of the data.
Audit mode to get started
Nothing is blocked during the pilot: you measure your real exposure before enforcing the policy.
A policy that fits you
Your markings, your internal identifiers, your approved assistants, your excluded sites. All in a single file.
Remote deployment
Installed across your entire Windows and Mac fleet via Intune, Group Policy, Jamf or Kandji. Users can neither remove it nor relax it.
A log your DPO can read
Data types and decisions, never the values. Nothing leaves the device.
FAQ
What people ask us most.
How much does it cost?
€6 excl. VAT (€7.20 incl. VAT) per user per month on the annual plan, €8 excl. VAT (€9.60 incl. VAT) on the monthly plan, after a 3-month trial with nothing charged. For more than 250 users, pricing is on quotation.
Does Tacite-IA itself use AI?
No. It relies on precise rules and checksum calculations (the check digits of an IBAN, a social security number or a bank card). The result: alerts that are explainable and instant, with no data sent off for analysis.
Do my messages go through your servers?
Never. Everything is analysed on the device, by the Tacite-IA program. Our server (data hosted in Europe) handles licences, updates and the administrator dashboard, which only receives data types and decisions, never content: no message or file ever passes through it.
What if it is a false alarm?
The user stays in control: “Send anyway” lets the message go, and the choice is logged. The rules can be tuned to your context.
Which tools does it work with?
Chrome and Edge: ChatGPT, Claude, Copilot and Microsoft 365 Copilot, Gemini, Le Chat, Perplexity and more, with an alert on every other site. For developers: Claude Code (terminal, VS Code, JetBrains), Cursor, Windsurf, OpenAI Codex, Gemini CLI and Copilot on the command line. Installed chat apps (ChatGPT, Windows Copilot, the Claude Chat tab) allow no control at all: we recommend blocking them and using these assistants in the browser.
Does it slow people down?
No. Analysing a message takes a few milliseconds. A message with no sensitive data goes through as usual, with no window at all.
Does it work in other languages?
Yes. Detection recognises personal data, national identifiers, health data and confidentiality markings in the main European languages, and the window is shown in each user’s language. Ideal for groups and overseas subsidiaries.
How is it installed?
Remotely, like the rest of your software: your IT department deploys the Tacite-IA program with Intune, Group Policy, Jamf or Kandji. It also installs the Chrome and Edge extension and the Claude Code module. Employees have nothing to do.
Who is it for?
Organisations that handle sensitive data and whose computers are managed with a deployment tool (Intune, Group Policy, Jamf, Kandji…). That is what makes it possible to enforce protection on every device.
Community
Let’s build it together. For everyone.
Tacite-IA grows thanks to the people who use it. We are looking for companies to deploy it, testers to report false alarms, developers, DPOs and lawyers, trainers and agencies willing to share their needs. We stand with employees, customers and their data: come and join us.
Companies and agencies: deploy it, and tell us what is missing
Testers: report false alarms and missed data
DPOs, lawyers, trainers: help us raise awareness more effectively
Or email us: support@simplymy.fr