Legal

Privacy statement

Updated 2026-10-24

Draft: this text gets a legal review before launch.

About this statement

Tideform (tideform.ai) is a service of [Bedrijfsnaam], [Rechtsvorm], registered with the Dutch Chamber of Commerce (KvK) under number [KvK-nummer], located at [Adres]. In this statement we call this "Tideform", "we" or "us".

Tideform runs an AI process audit for small and medium-sized businesses. An AI interviewer talks with employees about their recurring work. The company then receives a report with processes, hours and opportunities for automation.

This statement is for four groups of people:

Visitors of our website.
People who request a call through the form.
Admins at our clients (the person who arranges the audit for the company).
Employees of our clients who are interviewed by the AI interviewer.

Do you have a question about privacy? Email us at [E-mailadres privacy].

Who is responsible for your data?

The GDPR (in Dutch: AVG) separates a controller (who decides why and how data is used) from a processor (who processes data on behalf of someone else).

Website, call requests and client data: here Tideform is the controller.
Interviews with employees: here the employer (our client) is the controller. Tideform is the processor and only acts on the employer's instructions. We sign a data processing agreement with the employer for this. Are you an employee? Then your employer is your first point of contact. You may also email us; we will forward your question to your employer and help them answer it.

Website visitors

No cookies, no tracking. Our website uses no cookies and no analytics or tracking tools. We do not follow you and we do not build a profile of you.

Local storage in your browser. Your browser only stores your language choice (Dutch or English). For Tideform staff, the browser also stores an owner code so they can open their admin area. This data stays on your own device. You can delete it at any time in your browser settings. Because this storage is strictly necessary for a function you ask for yourself, no consent is required (article 11.7a of the Dutch Telecommunications Act).

Technical data. To show you a page, our server technically receives your IP address and data about your browser. This is needed to run and secure the website. Legal basis: legitimate interest (article 6(1)(f) GDPR), namely a secure and working website.

People who request a call

A company can request an intake call through the form on our website.

Which data: name, company name, email address, phone number (if you enter it), team size, your preferred time and your question.

What for: to plan the call, prepare for your question and contact you about a possible audit.

Legal basis: legitimate interest (article 6(1)(f) GDPR). Our interest is answering a business request that you make on behalf of your company. You can always object to this.

Retention: 12 months after your request. If your company becomes a client, the periods for client data below apply.

Admins at clients

When a company becomes a client, we create an audit. The admin receives a personal admin link. That link is the key to the admin area: there is no password. So keep the link safe and do not share it.

Which data: the contact details from the call request, data needed to carry out and invoice the assignment, and the messages we exchange with you.

What for: to perform the agreement with your company, stay in touch about the audit and send invoices.

Legal basis: legitimate interest (article 6(1)(f) GDPR), because the agreement is with your company and not with you personally. For our financial records there is a legal obligation (article 6(1)(c) GDPR).

Retention: as long as we work together and 12 months after that. Invoices and other records for our accounts are kept for 7 years, because Dutch tax law requires this (article 52 of the General Tax Act, AWR).

Employees who are interviewed

This part is about the interviews. Here your employer is responsible and Tideform is the processor.

How it works

Your employer enters your first name and, if they want, your role. Your employer then sends you a personal link.
Through that link you talk with an AI interviewer for about 20 minutes. You can use your voice or type, in Dutch or in English.
The interviewer says at the start that it is an AI. The interviewer is Claude, an AI model made by the company Anthropic.
The questions are about your recurring work tasks: how often, how long, which systems, who you get work from and pass it to, what takes time and what you would like to get rid of.
At the end you see a summary of your own tasks. You can change tasks, remove them or add one.
Only after you approve the summary does your interview count in the report. You can stop before approving. An interview that is not approved is not in the report.

Which data

your first name
your role (if your employer enters it)
the text of the conversation
structured data about your tasks (such as how often and how long)
the status of your interview and whether you approved it
a note, if you add one
if you choose to show a process: a recording of your screen and voice, with a short note

We receive your first name and role from your employer. Everything else comes from you, from the conversation.

Voice

Do you use your voice? Then our own server turns your speech into text. For this we use the open-source software Whisper (through whisper.cpp). So your audio does not go to another company. The audio file is deleted right after it has been turned into text.

The interviewer's questions can be read aloud. This happens in one of two ways:

with your browser's built-in voice: then nothing is sent;
with Google Cloud Text-to-Speech through Google's European endpoint: then we only send the text of the interviewer's question to Google, not your voice or your answers.

What your employer can and cannot see

In the admin area your employer sees your first name and the status of your interview: invited, in progress, awaiting approval or approved.
Your employer does not see the text of your conversation.
The report shows processes, hours and automation potential per role and per department, combined. Quotes are linked to a role, never to a name.
Nobody gets a score or a rating.

Please note: in a small team, a role can sometimes point to one person. If you do not want something from your conversation to appear as a quote in the report, change or remove it in your summary before you approve. If you want something removed after approval, contact your employer or us.

What we do not do

We do not recognise emotions. This is also prohibited at work (article 5(1)(f) AI Act).
We do not assess how well someone works.
No automated decisions are made about people (article 22 GDPR).
We do not use interviews to train AI models. Anthropic and Google do not use the content for training either (see below).

Legal basis

Your employer decides the legal basis. Usually this is legitimate interest (article 6(1)(f) GDPR): the company wants to know where time is lost and what it can automate. Consent is usually not a good legal basis at work, because employees are not truly free to say no.

Your employer must inform you about the audit beforehand. Does your company have a works council (in Dutch: ondernemingsraad, OR)? Then the works council may have a right of consent. This applies to rules on processing employees' personal data (article 27(1)(k) of the Dutch Works Councils Act, WOR) and to systems that are suitable for observing or checking employees' attendance, behaviour or performance (article 27(1)(l) WOR). Tideform is not meant to measure performance, but the employer must check for itself whether the works council has to consent.

Please do not share sensitive data in the conversation, such as about your health, religion or political views. It is not needed for the audit.

Retention

We keep interview data and screen recordings until 12 months after the last activity in the audit. Your employer can ask us to delete them earlier, and you can have a screen recording deleted yourself. We never keep interview audio longer than needed to turn it into text. Backups (daily snapshots of the storage) are kept at most 5 days longer.

Showing a process once

When we build an agent, your employer may ask you to demonstrate one process: you share your screen and talk through what you do. This is voluntary. You start and stop the recording, watch it back first and only then send it. Only Tideform and your employer can view it, and we only use it to build the agent for that process. You can always have it deleted, also yourself through the same link. Nobody watches outside that one recording.

People in system scan files

For the system scan, a client can upload exports and documents from its own systems, such as a list of invoices or quotes. These can contain names of customers, suppliers or colleagues. We process those files as a processor for the client, under the data processing agreement. Claude counts volumes and puts them next to the interviews; the report contains counts and ways of working, not data about people. The client can delete the files itself. Otherwise they are deleted together with the audit. The client can also admit us as a member of the company's own Claude workspace. There we only read the owner's mailbox and shared inboxes, never employees' personal mailboxes, and we only take over volumes and kinds of work, no content.

AI and the AI Act

You know you are talking to an AI. The interviewer says so at the start of the conversation. This is how we meet article 50(1) of the AI Act, Regulation (EU) 2024/1689.
No emotion recognition at work. Article 5(1)(f) of the AI Act prohibits AI systems that infer people's emotions at work. Tideform does not do this.
No assessment of people. Tideform is not meant to make decisions about employees or to assess their performance. Clients may not use it for that either.

Where your data is processed

Hosting: our server and database run at Fly.io in the Amsterdam region (EU).
AI processing: in production, Claude runs through Google Cloud Vertex AI in the EU multi-region. According to Google, the machine learning processing then stays within the EU. Google processes the data under its data processing terms. According to Google's documentation, prompts and responses for partner models such as Claude are not shared with third parties, including the partner. So through this route Anthropic does not receive the content directly.
No training: Google Cloud's terms say Google does not use customer data to train AI models without permission. Anthropic's commercial terms say Anthropic does not train models on customer content.
Test mode: during testing we use Anthropic's API directly. That route is not pinned to the EU. The app then clearly shows "test mode". Client audits run in production.
Speech to text: on our own server.
Text to speech: in your browser, or through the EU endpoint of Google Cloud Text-to-Speech.

Some of these companies are American. Where data may go to the United States, we use appropriate safeguards. That is the EU-US Data Privacy Framework (an adequacy decision of the European Commission) for companies certified under it, or the European Commission's standard contractual clauses. More details are in our data processing agreement.

We never sell your data. We only share it with:

Fly.io (hosting, Amsterdam region). Fly.io, Inc. is an American company and is certified under the EU-US Data Privacy Framework.
Google Cloud (Vertex AI for Claude in the EU, and Text-to-Speech through the EU endpoint). Google is certified under the Data Privacy Framework and also uses standard contractual clauses.
Anthropic (maker of Claude). Through Vertex AI, Anthropic does not receive the content directly. Only in test mode does content go to Anthropic directly. Anthropic uses standard contractual clauses for transfers outside the EU.
ElevenLabs (optional voice service). We do not use ElevenLabs by default. If we do use it for a client, we agree this in advance.
Your employer, as far as interviews are concerned: your first name, the status and the combined report. Not the text of your conversation.
Authorities if the law requires us to.

Security

We take appropriate technical and organisational measures (article 32 GDPR). For example: encrypted connections (HTTPS), personal links that cannot be guessed, storage in the EU, deleting audio right away, and access for as few people as possible. If you notice something that is not right, email us at [E-mailadres privacy].

Your rights

Under the GDPR you have these rights:

Access (article 15): you may know which data we hold about you.
Rectification (article 16): you may have incorrect data corrected.
Erasure (article 17): you may ask us to delete your data.
Restriction (article 18): you may ask us to pause the use of your data.
Portability (article 20): you may receive your data in a common file format, if the processing is based on consent or a contract.
Objection (article 21): you may object to processing based on legitimate interest.
No automated decisions (article 22): Tideform does not make decisions about people.

How to use your rights

Website, call requests, client data: email us at [E-mailadres privacy].
Interviews: contact your employer. If you email us, we forward your request to your employer and help them answer it.

We may ask you to show who you are, so we do not give data to the wrong person. We reply within one month. If your request is complex, this can take two more months; we will tell you so within the first month. A request is free of charge.

Making a complaint

Do you disagree with how we handle your data? Please tell us first, so we can find a solution together. You also always have the right to file a complaint with the Dutch Data Protection Authority, the Autoriteit Persoonsgegevens (article 77 GDPR).

Changes

We may change this statement, for example when our service or the law changes. The date of the latest version is at the top. For major changes we inform clients in advance.

Contact

[Bedrijfsnaam], [Rechtsvorm]
[Adres]
KvK: [KvK-nummer]
Email: [E-mailadres privacy]