---
title: "Privacy Policy — gptagent"
description: "How gptagent collects, uses, shares and protects personal data, and the rights you have under UK GDPR."
url: https://gptagent.ai/legal/privacy-policy/
lang: en
---

# Privacy Policy

**Version 1.2 — last updated 22 November 2026** (adds the sign-up details and the internal message to our team about a new sign-up)

> **Scope.** This policy covers personal data of **website visitors and enquirers**, and the sign-up details described in section 2.2 of people who **create an account** on our sign-up page. When gptagent processes data on behalf of a business customer (e.g. call recordings, transcripts, and the personal data of that customer’s callers), gptagent acts as a **processor** under the customer agreement and a **Data Processing Agreement (DPA)** — that processing is governed by those documents, not this policy.

## 1\. Who we are (Data Controller)

**UPSCLD LLP** (trading as “gptagent”), a limited liability partnership registered in England and Wales (LLP number **OC453110**), registered office **3rd Floor Suite, 207 Regent Street, London, England, W1B 3HH**, is the data controller for the personal data described in this policy.

-   Privacy contact: **[privacy@gptagent.ai](mailto:privacy@gptagent.ai)**
-   Data Protection Officer: **not applicable** (we are not required to appoint one)

We process personal data in accordance with the **UK GDPR** and the **Data Protection Act 2018**.

## 2\. Personal data we collect

-   **Enquiry & booking data** — your name, work email, and organisation, plus any details you choose to share about your inbound line so we can assess fit and respond to your enquiry: your **role**, **industry**, approximate **monthly volume**, **channel mix** (voice / chat), **current telephony/CRM**, and anything else you write to us.
-   **Usage & analytics data** — pages viewed, referring page, approximate (city-level) location, device and browser type, and interactions, collected via Google Analytics **only where you have consented** (if you have, a verified enquiry is also counted there with the industry, volume and channel options you selected — not your name, email or message). Separately, Cloudflare Web Analytics counts page views and page-load speed without cookies or other storage on your device.
-   **Chat data** — if you start a conversation in the chat window on our website: your messages, the identifiers the chat window then stores in your browser, the website host and, in voice mode, a recording of the call. Details in section 2.1.
-   **Sign-up details** — if you create an account: your device and browser type, the country and city of your IP address, where you came from and, only if you accepted analytics on our website, your visit path; and an internal message to our team when you confirm your email address. Details in section 2.2.
-   **Technical & log data** — IP address and request metadata processed by our hosting/CDN provider for delivery, security, and abuse prevention (including anti-spam checks on our forms).
-   **Cookies and similar technologies** — see our [Cookie Policy](https://gptagent.ai/legal/cookie-policy/).

We do **not** intentionally collect special-category data through the website. Please do not send us sensitive personal data via enquiry forms or the chat.

### 2.1 Chat on our website

Every page of gptagent.ai loads a chat window from widget.gptagent.ai. It runs on our own product, and UPSCLD LLP is the controller of the data in it. **You chat with an AI assistant.** You do not have to start a chat to use the site. Without your messages or voice, the assistant cannot answer you.

The chat window stores nothing in your browser until you send your first message or start a voice call. Then it stores a random identifier and the identifier of the conversation (see section 4.1 of our [Cookie Policy](https://gptagent.ai/legal/cookie-policy/)) and sends them to our servers.

**What we receive when you use the chat**

-   the messages you type, and the assistant’s replies;
-   in **voice mode** (the microphone button, only after you allow the microphone): your voice and the assistant’s voice;
-   the random browser identifier (`gptagent.widget.customer`, for example `anon-…`), which recognises the same browser across conversations, and the identifier of the current conversation. In our database we link each conversation to the browser identifier. These identifiers do not contain your name, but they are personal data. The browser identifier stays until you or your browser clear site storage; the conversation identifier stays until you close the tab;
-   the language of the chat window, the website host where the chat runs (for example `gptagent.ai`, not the full page address), and the start and end time of each conversation.

Our website does not send your name or email address when a chat starts. If you write or say them, or other personal data, they become part of the conversation record. Your IP address reaches our servers with each request. We use it to limit the number of requests. Our web access logs can contain your IP address and browser type (section 6). The IP address is not a field of the conversation record.

**How the assistant answers**

For a text reply, we send your message, the conversation so far and the assistant’s instructions to an AI language model. They go through our own AI model gateway (OKI-TOKI is our brand, not a separate company), which runs on our servers in Germany and does not log the text of requests. The gateway sends them directly to **OpenAI**; the model we use at present is **GPT-4o mini**. We also send each message you type through the same gateway to an OpenAI automatic moderation model. This check can stop an abusive message from being answered. The AI writes the replies; we do not use the chat to make decisions about you that have legal or similarly significant effects.

We do not train AI models on chat conversations. Under our terms with OpenAI and Google, they may not use this data to train their models.

In **voice mode**, our servers stream your voice to **Google** (Gemini Live API), which answers by voice. We record every voice call (your voice and the assistant’s voice) as an audio file, and we keep a written transcript of it. If you do not want to be recorded, use text chat.

**Talking to a person**

A person does not join the chat, and the assistant does not send your conversation to our team. If you want to talk to a person, the assistant gives you our [contact form](https://gptagent.ai/contact/) and email address. Our team can read the conversation record later, as described below.

**Who can see the conversation**

Members of our team with supervisor, administrator or auditor access to our console can read chat transcripts and listen to voice recordings.

**Where it is stored and for how long**

We store the conversation record (messages, identifiers, website host, times) in our database, and a copy of each conversation in a file, on servers of our hosting provider, **Hetzner**, in data centres in Germany. Voice recordings are audio files on these servers. While a conversation is active, we also hold it in a session store; it expires after 20 minutes without activity. Database backups contain the conversation records but not the voice recordings. Retention is described in section 6.

### 2.2 Sign-up details and the message to our team

You create an account on our sign-up page at app.gptagent.ai/signup. When you send the sign-up form, we record these details together with what you type in the form:

-   **device or operating system, and browser**, for example “iPhone · Safari” or “Windows · Chrome”. We work them out from the identification string (User-Agent) that your browser sends. We keep only these names in your sign-up record, not the string itself;
-   **country and city** of your IP address. Our own server looks them up in a geolocation database file that it holds (DB-IP Lite); your IP address is not sent to anyone for this lookup. We do not save your IP address in your sign-up record. Our servers and hosting provider process it as described under “Technical & log data” in section 2;
-   **where you came from**: the source tags in the address of the sign-up link (`utm_source`, `utm_medium`, `utm_campaign`, `utm_content`). For example, a link on our pricing page tells us that you came from the pricing page;
-   **your visit path**, only if you clicked **Accept** in the cookie banner of our website: the addresses of the last 15 pages you opened on our website in that tab, the host name of the website that sent you to the first of these pages, and the time of the first of these pages (section 4.2 of our [Cookie Policy](https://gptagent.ai/legal/cookie-policy/)). If you did not accept, we record only that you did not consent. The sign-up page keeps a copy of these details in the session storage of its tab (`gptagent.signupVisit`) until you close the tab, so that a reload of the sign-up page does not lose them.

We store these details in our database on Hetzner servers in Germany: first with your unconfirmed sign-up, then with your user account record when you confirm your email address.

**The message to our team.** When you confirm your email address, our server sends a message about the new account to a private channel of our team in **Discord**. The message contains:

-   the company name you entered, and the **domain** of your email address (for example `example.com`) — not the full address;
-   the plan (for example the free trial) and the language of the sign-up;
-   the country and city, and the device and browser type;
-   where you came from: the source tags and, if you accepted analytics on our website, the referring website, the time from your first recorded page to the confirmation of your email address, and your visit path; otherwise a note that you did not consent;
-   a link to your account in our internal console.

The message does not contain your full email address, your name (unless your company name contains it) or your IP address. We send no message if you do not confirm your email address. Members of our team who have access to the channel can read the message.

**Geolocation data.** The country and city come from [IP Geolocation by DB-IP](https://db-ip.com). We use the database as published; it is licensed under the [Creative Commons Attribution 4.0 International License](https://creativecommons.org/licenses/by/4.0/).

## 3\. Why we use it, and our lawful basis (UK GDPR Art. 6)

| Purpose | Lawful basis |
| --- | --- |
| Answer your questions in the chat, including by voice | **Legitimate interests** (answering questions about our product that visitors ask us) |
| Check chat messages for abuse and limit the number of requests | **Legitimate interests** (protecting visitors and the service) |
| Keep chat transcripts and voice recordings to check and improve the quality of the assistant’s answers, help our team follow up on questions asked in the chat, and deal with complaints, disputes and legal claims (not to train AI models) | **Legitimate interests** (making sure our assistant answers visitors correctly; protecting our legal position) |
| Respond to your enquiry, assess fit and follow up on it | Steps at your request prior to a contract / **legitimate interests** (responding to, and following up on, business enquiries you send us) |
| Record your sign-up details (device and browser, country and city, source) and tell our team about a new confirmed account in an internal Discord message | **Legitimate interests** (knowing promptly about new customers so that we can help them start, and understanding which sources bring new customers). You can object at any time at **[privacy@gptagent.ai](mailto:privacy@gptagent.ai)** |
| Record your visit path on our website, store it with your sign-up details and show it in the internal Discord message | **Consent** (you click **Accept** in the cookie banner; Cookie Policy, section 4.2). To withdraw consent for a path already sent with a sign-up, email **[privacy@gptagent.ai](mailto:privacy@gptagent.ai)** |
| Operate, secure, and improve the website | **Legitimate interests** (running and protecting our site) |
| Analytics and audience measurement with Google Analytics | **Consent** (for non-essential/analytics cookies) |
| Cookieless page-view and performance statistics (Cloudflare Web Analytics) | **Legitimate interests** (understanding and improving our site) |
| Wider marketing communications | **Consent** |
| Comply with legal obligations; establish/defend legal claims | **Legal obligation** / **legitimate interests** |

Where we rely on legitimate interests, we have assessed that they are not overridden by your rights. Where we rely on consent, you may withdraw it at any time (this does not affect earlier processing). You can ask us to stop contacting you at any point by emailing **[privacy@gptagent.ai](mailto:privacy@gptagent.ai)**.

## 4\. Who we share it with (processors & recipients)

We use trusted third parties who process personal data on our behalf under appropriate contracts, including:

-   **Cloudflare** — website hosting, CDN, security, privacy-friendly analytics, and form anti-spam (Turnstile);
-   **Google** — Google Analytics (GA4) and Search Console; in chat voice mode, the Gemini Live API (section 2.1);
-   **Hetzner** — hosting, in Germany, of the servers that run the chat, our AI model gateway, and store conversation records and voice recordings, and of our backups;
-   **OpenAI** — the AI language model that writes the chat assistant’s text replies, and the model that checks typed messages for abuse (section 2.1);
-   **our email and enquiry-handling tools** — receiving and responding to enquiries;
-   professional advisers, and authorities where required by law.

We also use **Discord** (Discord Inc., United States), under its own terms, to deliver and keep the internal messages to our team about new confirmed accounts (section 2.2).

We do not sell your personal data.

## 5\. International transfers

Some of our providers are located outside the UK (including the United States). The servers that run the chat and store conversation records, and our backups, are in Germany (Hetzner); the UK recognises the EEA as providing adequate protection. Chat messages sent to OpenAI, and voice-mode audio sent to Google, may be processed outside the UK, including in the United States. The internal sign-up messages sent to Discord are processed in the United States. Where personal data is transferred outside the UK, we rely on appropriate safeguards under UK data-protection law — such as **UK adequacy regulations**, the **International Data Transfer Agreement (IDTA)**, or the **EU Standard Contractual Clauses with the UK Addendum** — together with additional measures where needed. You can ask for information about these safeguards, or a copy of them, at **[privacy@gptagent.ai](mailto:privacy@gptagent.ai)**.

## 6\. Retention

We keep personal data only as long as necessary for the purposes above:

-   enquiry/booking data — for up to **14 months** after our last contact, unless a longer period is required (e.g. for legal/accounting reasons);
-   analytics data — for the retention period configured in the analytics tool (**14 months** in GA4);
-   chat conversation records in our database (messages, transcripts, identifiers, and the link between a conversation and your browser identifier) — an hourly job deletes them once they are older than **14 months**, counted from the end of the conversation (or its start, if no end time is recorded). This does not remove the identifier from your browser (section 2.1);
-   chat voice recordings and conversation files — a job that runs every minute deletes them once they are more than **30 days** old. Deleting a recording does not delete the transcript in our database;
-   active chat session data — expires after **20 minutes** without activity;
-   database backups (which include chat records, but not voice recordings or conversation files) — our backup policy keeps **14 daily** backups, so a deleted chat record can remain in a backup until that backup is removed;
-   sign-up details (section 2.2) — with your user account record, while the account exists. If you do not confirm your email address within **48 hours**, the sign-up expires; the unconfirmed record stays in our database until you sign up again with the same email address or we delete it;
-   internal sign-up messages in Discord — Discord keeps each message until we delete it; it is not deleted automatically. We delete the message about your account if you ask us to erase your data (section 7);
-   log data — for a short period for security and troubleshooting; our central log store has a **14-day** retention setting.

## 7\. Your rights

Under UK GDPR you have the right to: **access** your data; request **rectification**; request **erasure**; **restrict** or **object** to processing; request **portability**; and **withdraw consent**. To exercise any right, contact **[privacy@gptagent.ai](mailto:privacy@gptagent.ai)**. We will respond within the statutory time limit and may need to verify your identity.

You also have the right to complain to the UK supervisory authority, the **Information Commissioner’s Office (ICO)** — ico.org.uk — though we’d appreciate the chance to address your concern first.

## 8\. Security

We use appropriate technical and organisational measures to protect personal data. No method of transmission or storage is completely secure, but we work to protect your information and to notify you and the ICO of any qualifying breach as required.

## 9\. Children

The website is intended for business users and is not directed at children under 18. We do not knowingly collect their personal data.

## 10\. Changes

We may update this policy from time to time; the “Last updated” date reflects the current version. Material changes will be highlighted where appropriate.

## 11\. Contact

Data protection queries: **[privacy@gptagent.ai](mailto:privacy@gptagent.ai)** — **UPSCLD LLP**, 3rd Floor Suite, 207 Regent Street, London, England, W1B 3HH.
