By default, it stays on one computer.

Before anything is recorded

Every person in the conversation reads and signs their own consent form. The conversation is only recorded if everyone agrees. Nothing on the form is ticked in advance.

The standard route

This applies to every conversation.

1. You record the conversation and upload the file through a private link. Your files go into a folder under your name that nobody else can see. I download them to my own computer and delete them from the upload service within 48 hours.

2. Speech-recognition software on my own computer turns the audio into a transcript. The audio is not uploaded to any online service.

3. The transcript is analysed on that same computer, against fixed, written criteria, with a language model that also runs locally.

4. I review the results and the report before it is sent to you.

Some recorders keep a copy online: cloud recording in Zoom or Teams, AI notetakers, or automatic backup of voice memos to iCloud or Google. That copy falls outside this process. To avoid it, use a recorder that saves only on your device, or switch backup off for the recording.

Option A: extended analysis with Claude

Optional

What you gain: richer and more detailed feedback than the local analysis alone. It also helps me improve the feedback criteria. The local analysis is always done, and is useful on its own.

How it works: I also analyse the transcript with Claude, an AI model made by the US company Anthropic, using my own paid Claude Pro account.

What I do to limit the risks

• I replace the names of people and companies with codes before uploading. The audio is never uploaded.
• I keep the setting that lets Anthropic use conversations to train its models switched off.
• I delete the conversation from my Claude account within 2 weeks after the report is delivered. According to Anthropic's current privacy policy, it is then removed from its systems within 30 days.
• I protect the account with a strong password and two-step login.

The risks you should know about

• No contract about your data. Claude Pro is a consumer service. Anthropic handles the transcript under its own consumer terms and privacy policy, not on my instructions.
• Terms can change. I can only promise what the current terms say and what I control myself.
• Processing in the United States, where data protection rules differ from the EU's. This transfer relies on your explicit consent.
• Longer retention in some cases. Conversations flagged by Anthropic's safety systems can be kept longer than 30 days.
• Codes are not full anonymity. They reduce the chance that you can be recognized from the content, but do not remove it.

Option A is only used if everyone in the conversation says yes.

Option B: keeping the transcript

Optional

What you gain: if you submit more conversations later, I can show you how they develop over time.

How it works: I keep the transcript, with names of people and companies replaced by codes, for two purposes only: to test and improve the feedback criteria, and to compare it with your later conversations. It is never shared, published or quoted in a way that could identify anyone, and it is deleted no later than 24 months after your last submitted conversation.

Option B is only used if everyone in the conversation says yes. You can withdraw it at any time, and the transcript is then deleted.

Who sees what, and for how long

In a group program, such as the ScaleUpNation Board Program, anonymized patterns from all conversations may also be discussed with the group: no names, no companies, no quotes that could identify anyone. The program's consent form describes this.

Your rights

Your consent is voluntary. You can withdraw it at any time by emailing jp@jpgrolle.com; data processed up to that point is then deleted and not used further. You can ask to see, correct or delete your data. If you have a complaint, you can contact the Autoriteit Persoonsgegevens.

I treat every conversation as confidential, in line with the ICF Code of Ethics.