Namarix
PrivacyTermsImpressumDatenschutz

Privacy Policy

Last updated: 9 November 2026

This Privacy Policy explains how Young Hoon Lim, operating under the Namarix brand ("Namarix", "we", "us", or "our"), collects, uses, shares, and protects personal data when you visit our website, communicate with us, use Namarix, or interact with a Namarix-powered AI agent.

Namarix is a brand name, not a separate legal entity. The business is a sole proprietorship registered in Lüneburg, Germany. Namarix is also the product name used for our AI service, which is sold only under a private contract and has no public self-serve signup. If a GmbH is formed and becomes the operator of Namarix, this Privacy Policy will be updated.

1. Controller

Controller for this website and Namarix's own business processing:

Young Hoon Lim Operating under the Namarix brand Hasenburger Ring 51d 21335 Lüneburg Germany Email: mark@namarix.ai Website: https://namarix.ai

VAT ID: DE458015605

No separate Data Protection Officer has been appointed. Privacy requests can be sent to mark@namarix.ai.

2. Controller / Processor Roles

For Namarix's own website, sales, marketing, account administration, product administration, security, billing, legal, and business communications, Young Hoon Lim is the controller.

For workspaces that a client operates, and for the free workspace a podcast partner operates in its own account, that client or podcast partner is the controller for the data collected there. Young Hoon Lim / Namarix acts as a processor under its documented instructions and a data processing agreement, and its own privacy notice is the one that governs. We never sell or license data from those workspaces, and we never use it for the company-level market insights or the reports to podcasts with an Audience Partnership described in section 4. Usage statistics from client workspaces. Unless the client has turned it off in its Data controls settings, we calculate a fixed set of weekly usage figures from the workspaces our clients run, on the client's instruction and as its processor. The figures are: sessions, sessions with 10 or more turns, answers, email captures, citation clicks, median turns per session, email capture rate and the rate of sessions with 10 or more turns. They are combined across all client workspaces where this is on and calculated per week and per channel (website and Slack). Any figure based on fewer than ten visitors is dropped. The figures contain no name, email address, company, question text, link or workspace identity, we keep no separate copy of your data to calculate them, and once calculated they are not personal data. We use them only to improve our own service and internal systems, and we show a figure outside Namarix only where it combines at least three clients. We never use data from client workspaces for our own sales or marketing, or to find, score, target or contact you or your company on our own behalf.

For workspaces that Namarix operates itself, such as topic workspaces built from episodes that podcasts have allowed us to index and our own content library, there is no client: Young Hoon Lim is the controller, and this Privacy Policy is the notice that applies to you. A podcast whose episodes appear in such a workspace does not operate it and does not decide what is collected. This stays true where we send that workspace's newsletter in the podcast's name (section 7a): the podcast does not receive your email address. If it has an Audience Partnership with us, it receives the weekly company-level report described in section 4, which is designed to identify no person and never contains names, email addresses, contact details or conversations. If any part of a report could still relate to a person, the podcast holds that part as a separate controller, bound by our terms not to identify anyone, not to sell the report and to share it only as section 4 describes.

Where Namarix independently decides why and how personal data is used, such as for its own account administration, security, billing, legal compliance, and legally permitted marketing, Namarix acts as an independent controller for that processing.

3. Personal Data We Collect

We may collect:

  • Administrator account data (our clients' staff, not website visitors): email address, sign-in method, workspace name and settings.
  • Communications data: messages, demo requests, support requests, contact-form submissions, preferences, and opt-out status (see section 16 for demo requests).
  • Newsletter data: your email address, the workspace whose newsletter you asked for, the status of your subscription, the records of your sign-up, confirmation and unsubscribe, and a log of what we sent. See section 7a.
  • AI agent conversation data: the work email address you choose to submit, every message in the conversation, timestamps, a session identifier we generate, and, derived from the conversation by an AI model, a short summary plus the main problem raised, any objection, the solution discussed and language detected. We do not ask you for your name, job title, seniority or department, and we do not store them. Community members may share their own job title and company when they introduce themselves; we store that as they wrote it and never look it up.
  • Client content and configuration data: the webinar, podcast, video, transcript and knowledge-base material a client uploads, together with speaker names and profile links, plus brand, prompt and agent configuration.
  • Security data: when the AI agent makes requests to our servers, your IP address, the session identifier and the event type are recorded so we can detect abuse, rate-limit requests and investigate incidents. The reports of clicks and actions described in section 5a are excluded, except the reports about the invitation to the community room. These records are kept for 180 days and then deleted.
  • Widget storage on your device: a conversation thread held in your browser's local storage, described in section 8. We set no analytics or advertising cookies and run no third-party tracking scripts. The measurement described in section 5a stores nothing on your device.
  • Embedding-site data: the web address of the page a client has installed the agent on, recorded so we can confirm the agent is only running where the client authorised it.
  • Company information derived from your email domain: company name, industry, sub-industry, headquarters city, approximate employee range, product line and typical customers, gathered from publicly available web sources. We do not access private contact databases and we do not look up you personally. See section 5.
  • AI agent usage data: clicks on sources in answers and other actions after an answer, for example the display, closing and submitting of the email card, a click on a starter suggestion, the display and click of the invitation to our community room, and the request to be contacted in Slack, each with the event type, the time and the identifiers named in section 5a. Details in section 5a.
  • Source-of-visit data: the campaign parameters in the address you opened (utm_source, utm_medium, utm_campaign, utm_content), the domain name of the referring site and the path of the page you opened. If you request a demo through our website, we also store the campaign parameters of the page you came from.
  • Topic and company assignment: one topic category per question from a fixed list, and the assignment of your conversation to a company based on the domain of your email address. Details in section 5a.

Do not submit sensitive personal data, health data, financial account data, government ID data, special-category data, confidential information, or trade secrets into the public AI agent unless we have expressly agreed in writing to handle that data.

4. How We Use Personal Data

We use personal data to:

  • provide, operate, secure, maintain, and improve the website, the Namarix service, the AI agents and the admin dashboards;
  • create and manage workspaces, accounts, content pages, sessions, notifications, and client configurations;
  • let visitors interact with an AI agent on a client's content page;
  • record, analyze, summarize, and route AI-agent conversation history;
  • deliver results to the client operating that agent, which may be a HubSpot contact and note, a row in the client's Google Sheet, a Slack message, or a webhook the client configured;
  • tailor AI-agent responses to a visitor's company domain or business context;
  • improve retrieval quality, prompts, evaluation and product quality using aggregated, de-identified learnings;
  • conduct Namarix sales, marketing, and customer research where legally permitted;
  • produce and license aggregated, company-level market insights, and prepare weekly company-level reports for podcasts with an Audience Partnership, from interactions with workspaces we operate ourselves, as described below;
  • evaluate which sources in AI agent answers are opened and which actions follow an answer (section 5a), to judge the quality of answers and citations, to give clients and podcast partners aggregated reports without names, email addresses or question text, and to classify enquiries for internal follow-up;
  • respond to enquiries, provide support, manage contracts and invoicing, prevent misuse, maintain security, comply with law, and enforce agreements.

We do not disclose one client's identifiable prospect data or confidential content to another client.

Company-level market insights

From interactions with workspaces that we operate ourselves (see section 2), we produce aggregated, company-level market insights. These insights describe organisations, not people: for example, that an organisation with at least 50 employees showed interest in a topic during a given week, which cloud platform or existing provider was mentioned, and an estimated stage of evaluation. We license these insights to businesses such as cloud provider partner teams, software vendors and partner-tooling vendors. We only release an insight about an organisation when at least three different people from that organisation contributed to it. Insights are released in weekly batches, only after the week has ended, and use fixed labels, never the words of a question. We never release names, email addresses, job titles, the text of your questions or anything else that identifies you, and we exclude sole traders, freelancers and organisations with fewer than 50 employees. Recipients are contractually prohibited from trying to identify any individual. Podcasts with an Audience Partnership receive a different weekly report, described in the next paragraph. We use only interactions that took place after this section was published, and we never use interactions with workspaces that a client or a podcast partner operates for this purpose.

Reports to podcasts with an Audience Partnership. We no longer accept new Audience Partnerships; this paragraph applies to podcasts whose Audience Partnership began before 9 November 2026. A podcast that allows us to index its episodes under an Audience Partnership receives a weekly report about the questions asked about its episodes in workspaces we operate ourselves. The report describes companies, not people. It can name a company whose employee asked about the podcast's episodes, where that person gave a work email address, together with the company's size, industry and headquarters, a short summary of the topic of each question and of the problems and objections raised, and totals such as visitors, conversations, the podcast's most-used episodes and how often its episodes were cited and their links clicked. It never contains your name, email address, job title or other contact details, a full conversation or the words of your question. We leave out sole traders and freelancers, anyone who has objected to this use, and any organisation that has asked to be left out. The podcast may use the report to run and grow its show and its own business, and may share it with its sponsors, guests and listeners only at company level, as a company name and a summarised topic, and only for companies with at least 50 employees. Our terms prohibit it from trying to identify anyone, combining the report with other information to do so, contacting anyone about what they asked, and selling or licensing the report. Where a company has at least 50 employees and at least three people from it asked questions, a report entry does not identify any one of them and we treat it as aggregate information. Where that is not the case, an entry could relate to you, and the report relies on legitimate interest (Art. 6(1)(f) GDPR); you can object as described in section 13. We use only interactions that took place after this paragraph was published, and never interactions with workspaces that a client or a podcast partner operates.

5. AI Agent And Email Capture

When you interact with a Namarix-powered AI agent, the agent tells you at the start of the conversation that it is AI. We store the content of your conversation, and, if you choose to submit it, your work email address. The conversation record includes the questions you ask, the business problems you describe, objections you raise, the topics you are interested in, and how long you engaged.

Submitting a work email is optional and is asked for once per conversation, together with the offer to subscribe to the workspace's weekly newsletter. We use the address so the agent can tailor its answers to your company context, for the newsletter once you have confirmed your subscription (section 7a), and, where the workspace is operated by one of our clients, so that client can respond to your enquiry and the conversation can be connected to that client's CRM where the client has enabled one. On a workspace Namarix operates itself there is no client, and nothing is passed to a client.

Company look-up from your email domain. After you submit a work email, we take the domain part of the address (for example „example.com“, never the part before the @) and run a public web search on it. Where the search result alone is too thin, we also fetch the published content of the public company page that the search points to. We then use an AI model to summarise what was found into a short company profile: company name, industry, sub-industry, headquarters city, approximate employee range, product line, and typical customers. Only the domain and the address of a public web page leave our systems for this step. The two providers we use for it are named in our Sub-processor Register. This describes the organisation, not you personally. We do not buy or query personal contact databases, and we do not look up your name, job title, seniority, phone number or social profiles. Free and disposable email domains are excluded from this look-up entirely. The profile is stored with your conversation. If the workspace is operated by one of our clients, it is included in what we pass to that client, as described below. On a workspace that Namarix operates itself, such as GTM Intelligence, there is no client: Namarix is the controller and keeps the profile. We do not pass it to anyone as a lead record, and the only onward sharing is the company-level insights and reports described in section 4, which identify no person. Separately, if your question is one the client should answer directly, we send the text of that question, shortened to at most 160 characters, to the same web-search provider, scoped to the website of the workspace operator (the client, or Namarix on a workspace we operate), so we can point you at the right page. Nothing else from your conversation is sent there. Our legal basis for both is our legitimate interest and, where a client operates the workspace, the client's legitimate interest in answering your enquiry properly (Art. 6(1)(f) GDPR), and you can object to it at any time using the contact details below.

Checking the email domain. As soon as you have typed a well-formed address into the input field of the email card, and even if you then do not submit it, your browser checks whether the domain, only the part after the "@" and never the part before it, can receive email. For this your browser queries the public DNS service of Cloudflare (cloudflare-dns.com); Cloudflare receives your IP address and the domain. When you submit, our server repeats the same check through Cloudflare and, if that does not answer, through Google's DNS service (dns.google); in that case these services receive our server's IP address and the domain, not your IP address. If a service cannot be reached, your input is accepted anyway. Nothing is stored on your device. The legal basis is our legitimate interest in deliverable addresses and in avoiding typing errors (Art. 6(1)(f) GDPR).

If the AI agent is operated for one of our clients, your email, your conversation, a short AI-generated summary of it, and the company profile above are passed to that client so they can respond to your enquiry and understand what their audience is asking. We also give the workspace operator reports on activity in its workspace. A report can name the company or work domain behind a visitor and give a short summary of the topics asked about. The workspace operator may share those company-level topic summaries with its partners, such as sponsors, guests and clients, and with its own team. Neither we nor the workspace operator may share your name, your email address or the words of your question in this way. Where a company has at least 50 employees and at least three people from it asked questions, a company-level summary does not identify any one of them and we treat it as aggregate information. Where that is not the case, the sharing relies on legitimate interest (Art. 6(1)(f) GDPR) and you can object to it as described in section 13. Unless the client has turned it off, we include use of its workspace in the weekly usage figures described in section 2. We never share one client's identifiable prospect data with another client. If the client has enabled it, the client also receives the summary of opened sources described in section 5a. On a workspace we operate ourselves, your email, your conversation and the company profile are not passed to any client or podcast.

We only write to subscribers. The email card lets you ask for the workspace's weekly newsletter. After you submit, you first receive only an email asking you to confirm; we send the newsletter only after you confirm (double opt-in, section 7a). If you do not confirm, you receive no newsletter. Your consent covers that workspace's newsletter only, not other marketing and not sales outreach. Separately from the subscription, we process the address you submitted as this section describes, that is for the company look-up and, on a client's workspace, for passing it to that client; this applies even if you never confirm or later unsubscribe. Where a client operates the assistant, we send the newsletter in that client's name and on its behalf; any other reply or follow-up comes from that client, under its own privacy notice and its own responsibility. We send marketing messages only as sections 7 and 7a allow. The summary described in section 5a serves internal follow-up only. No message is sent to you on the basis of that summary, and it does not turn your address into a mailing list.

The AI agent does not make decisions that produce legal effects or similarly significant effects about you. AI-generated outputs may be inaccurate and should be reviewed before important decisions are made.

5a. Measuring Clicks And Actions In The AI Agent

What we measure. We measure, on our own servers, what happens after an answer from the AI agent. For this we use no third-party analytics services, cookies, pixels or scripts, and we do not pass this data to advertising platforms.

Clicks on sources. We record clicks on a source in the source list, the opening of a row in the "Reviewed sources" area, clicks on a link in the answer text and clicks on the source icon on a paragraph. The same applies to a click on a source in Slack: the link there first goes through our server, which records the click and forwards you immediately to the source.

Actions after an answer. We record the display of the email card, closing the email card, submitting the email card, clicking a starter suggestion, the display of the invitation to our community room, the click on the invitation to our community room, and your request to be contacted in Slack. That you joined the community room is recorded as described in the privacy information for our community room; we do not link it to your session in the AI agent, only at company level through the domain.

What is stored per event. We store an internal record identifier, a random event identifier to prevent double counting, the event type, the surface, the time of the action and the time the server received it, the identifiers of the workspace, the conversation and the answer and, for a click on a source, the identifier of the source. For a starter suggestion we store its position. Every answer belongs to a conversation. The events are attached to that conversation and, if you gave your email address, can therefore be linked to it. We therefore treat them as personal data.

What we do not do. In the event record we store neither your IP address nor details about your browser; the browser identifier is checked only momentarily to filter out automated access and is not stored. For reports about the community invitation, the security log described in section 3 also applies. We set no cookie for the measurement, write nothing to your browser's storage, and read neither your visitor identifier nor any other entry of your browser. The identifiers of the answer and of the source exist only in the memory of the open page. Answers restored from the saved conversation on a later visit are not measured. We do not fingerprint your device, do not record mouse movements and do not infer moods or characteristics from clicks. We do not use this data to train AI models.

Source of the visit. When you open one of our workspace or episode pages, our servers read the campaign parameters in the address (utm_source, utm_medium, utm_campaign, utm_content), the domain name of the referring site, only the host name and not the full address, and the path of the page opened without parameters. These details are stored with your conversation at your first exchange with the AI agent so that we can see through which channel conversations arise. We store no personal parameters and no identifiers issued to individual recipients.

Companies and topics. If you have given your email address, we assign your conversation to a company record that we find only through the full domain of your email address, by exact match, without matching names and without buying data; free and disposable domains are excluded. After the answer, an AI model assigns each of your questions to one of currently 20 fixed topic categories, or to none. For this, the provider TypeSafe AI (USA) receives the text of your question, without your email address and without details about your company. The assignment is used only internally, is not shown to you and is not used for decisions that affect you legally or in a similarly significant way.

Signal records and reports. From these details we build signal records: entries that record that a company, not a person, was interested in a topic. Reports of this measurement to clients and podcast partners contain only counts per episode, topic, source and day, without names, email addresses and question text, and show no values below the minimum thresholds published for them. Company-level insights and the weekly reports to podcasts with an Audience Partnership are passed to third parties only under the rules in section 4 and only from workspaces that we operate ourselves.

Invitation to our community room. If a community room is set up for an agent, the agent shows an invitation after an answer. We record on your conversation whether and when the invitation was shown and whether and when you clicked it, once each. In a test (A/B), your conversation is randomly assigned one of two variants based on the conversation identifier, with the invitation shown or without it; we store the variant to compare the effect of the invitation. The link goes through our server, an address containing "/join/", which forwards you to Slack without further details and stores or passes on nothing about you.

Summary for internal follow-up. If you gave your email address in a conversation, a short summary can be created from your source clicks and written to the CRM (HubSpot) of the operator of the agent. It serves internal follow-up only. For each episode it contains the episode title and the number of sources opened. It does not contain the text of your questions, topic categories, times, IP addresses or the addresses opened. Neither Namarix nor the operator sends you a message on the basis of this summary. For agents we operate ourselves, we are the controller for this. For clients' agents we write the summary only if the client has expressly switched this on; the client is then the controller and we process it on its behalf. The feature is off by default. The legal basis is Art. 6(1)(f) GDPR. You can object to this processing at any time (Art. 21(1) and (2) GDPR); we then delete the summary in the CRM and write no new one. More in section 13.

Retention. We keep the events described in this section for 90 days; after that only daily counts remain, with no link to a conversation or a person. Source-of-visit, topic and company assignments we keep like the conversation they belong to. If a conversation is deleted, the related events and signal records are deleted with it.

6. Legal Bases

For individuals in the EEA, United Kingdom, or Switzerland, we rely on the following legal bases where applicable:

  • Contract: to provide services, accounts, support, and requested product functionality.
  • Legitimate interests: to operate, secure, analyze, improve, and market our B2B services; understand buyer intent and product-market signals; prevent abuse; support client sales and marketing workflows; and improve product performance, provided these interests are not overridden by your rights and freedoms.
  • Legitimate interests (company-level market insights): producing and licensing aggregated, company-level market insights from workspaces we operate ourselves, which serves our interest and the interest of the recipients in understanding demand in B2B markets (Art. 6(1)(f) GDPR). We have balanced these interests against yours. You can request a summary of that assessment at mark@namarix.ai. Legitimate interests (company-level topic summaries): preparing reports for the workspace operator and enabling it to share company-level topic summaries with its partners, which serves the operator's and its partners' interest in understanding what organisations are interested in (Art. 6(1)(f) GDPR). Where the threshold in section 5 is met the summaries are aggregate information; below it, you can object as described in section 13. Legitimate interests (reports to podcasts with an Audience Partnership): preparing the weekly company-level reports described in section 4 and giving them to podcasts with an Audience Partnership, which serves our interest in crediting and keeping the podcasts whose episodes make our workspaces useful, and their interest in understanding which companies their episodes reach (Art. 6(1)(f) GDPR). We have balanced these interests against yours, and you can request a summary of that assessment at mark@namarix.ai. Where the threshold in section 4 is met the report entries are aggregate information; below it, you can object as described in section 13.
  • Legitimate interests (measurement and follow-up): the measurement of clicks and actions, the source, topic and company assignment and the summary for internal follow-up (section 5a) and the check of the email domain (section 5) rest on Art. 6(1)(f) GDPR. Our interest: to understand which sources and answers help users, to give clients and partners aggregated evaluations, to classify enquiries properly and to avoid typing errors in email addresses.
  • Consent: in particular for a workspace's newsletter, including choosing its recommendations from your questions (section 7a), and where we otherwise ask for consent, such as for certain marketing communications, optional cookies, or specific processing activities.
  • Legal obligation: to comply with accounting, tax, corporate, regulatory, security, and legal obligations.

You may object at any time to processing based on legitimate interests on grounds relating to your particular situation (Art. 21(1) GDPR), in particular to the company look-up under section 5, the measurement and the summary under section 5a, the reports to podcasts with an Audience Partnership under section 4, and to marketing and profiling activities. Where processing serves direct marketing, including profiling connected with it, you may object at any time without giving reasons (Art. 21(2) GDPR). Write to mark@namarix.ai; section 13 describes the procedure.

7. Marketing Communications

We may send B2B marketing communications only where permitted by applicable law. A workspace's newsletter is also governed by section 7a.

In Germany and much of the EU, marketing emails usually require prior consent unless a narrow legal exception applies. We treat cold outreach to Germany/EU prospects as a separate compliance-controlled workflow.

In the United States, commercial emails must identify the sender and include an opt-out method. Marketing emails will identify the sender, include a valid postal address where required, and provide an unsubscribe or opt-out method.

You can opt out of marketing communications at any time by using the unsubscribe link in an email or contacting mark@namarix.ai.

7a. A Workspace's Weekly Newsletter

What you subscribe to. The email card in the AI agent lets you subscribe to the weekly newsletter of the workspace you are asking questions in. It recommends episodes from that workspace's library. The picks are based on the questions you asked in that workspace. We do not use questions from other workspaces, your company profile or other people's questions for this. An AI system picks the episodes and writes the text; individual issues are not reviewed by a person before they are sent. The newsletter never repeats the words of your questions.

Who is responsible, and whose name it comes in. On a workspace a client operates, the client is the controller; we send the newsletter in its name as its processor, and the client's own privacy notice also applies. On workspaces we operate ourselves (section 2), we are the controller. Every email states who sends it and on whose behalf.

Confirmation (double opt-in). After you submit, we email you a confirmation link. The link opens a page where you confirm with one click; merely opening the link, for example by an automatic link scanner in your email system, confirms nothing. The link is valid for 7 days. That email contains no advertising. If you do not confirm, we delete the request after 30 days.

What we keep. To prove your consent we keep the workspace, your email address, where you signed up (for example the email card), the language, the version and a checksum of the consent text you saw, and the time, IP address and browser identifier (user agent) of your sign-up and of your confirmation; if you unsubscribe, when, how and, if you give one, why. We also log which issue went to you when, which episodes it recommended and whether it was delivered. We do not track whether you open an issue or which links you click in it; links lead to our pages without personal parameters.

Sending service. We send the newsletter and the confirmation email through SMTP.com (J2 Global Canada, Inc., a Ziff Davis company) from a Namarix domain (currently namarix.ai, with namarixhq.com and news.namarix.ai), using an address that does not accept replies. SMTP.com receives your email address and the content of each email, and by its own account stores sent messages for up to one year. For transfers outside the EU, see section 10.

Legal basis. Your consent (Art. 6(1)(a) GDPR and § 7(2) of the German Unfair Competition Act, UWG), which covers choosing the recommendations from your questions. We keep the proof records after your subscription ends on our legitimate interest in being able to prove consent and defend claims (Art. 6(1)(f) with Art. 7(1) GDPR), and the suppression list on our legitimate interest in not emailing you again after you unsubscribe (Art. 6(1)(f) GDPR).

Unsubscribing. Every issue has an unsubscribe link, and your email app can unsubscribe you with one click. It takes effect immediately for that workspace; on the unsubscribe page you can also unsubscribe from every newsletter sent through Namarix. You can also write to mark@namarix.ai. Withdrawing consent does not affect the lawfulness of what was sent before. Unsubscribing does not change your conversation or the other processing in section 5; your rights in section 13 cover those.

Suppression list. After you unsubscribe, after an address permanently bounces, or after a spam complaint, we add the address to a suppression list (address, scope, reason, date) so that you receive no further email. Bounces, spam complaints and unsubscribing from every newsletter apply to all email sent through Namarix; we are the controller of that shared list, also for client workspaces.

How long. Subscription data while you are subscribed. Unconfirmed requests: 30 days. Records of sign-up, confirmation and unsubscribe: until 3 years after the end of the calendar year in which the subscription ends. Send log: 12 months. Suppression list: as long as needed to honour your unsubscribe.

8. Cookies And Tracking

This website and the Namarix AI agent use no third-party analytics tools, no advertising pixels, no retargeting, no session replay, and no tracking cookies. The only cookies we set are strictly necessary ones: for signed-in administrators of the product (a signed session cookie and the authentication provider's cookies), and the AI agent's acceptance cookie described below. Our own measurement under section 5a runs on our servers and stores nothing on your device. Because no non-essential cookies are loaded, no consent banner is shown.

The AI agent widget stores your conversation in your own browser's local storage so the thread survives a page reload, under a key beginning „kc_thread_“. That entry holds a randomly generated session id, your messages, and, if you submitted one, your email and the company profile. It is deleted automatically after 30 days without activity, and immediately when you press Restart. One further entry („kc_active_“) is used only to stop two browser tabs writing to the same conversation. When a source citation is shown, your browser loads that source's site icon from Google's public favicon service, which means Google sees your IP address for that request. If we ever introduce analytics or marketing cookies, we will ask for your consent first and add a preference control to the site. See our Cookie Policy at https://namarix.ai/cookies for the full breakdown.

When you accept the terms in our AI agent, it saves that acceptance in your browser's local storage and in a first-party cookie, both named beginning with "kc_consent_", so you are not asked again on later visits. Together with the acceptance it stores a randomly generated visitor identifier. We use this identifier only to recognise you as the same visitor of the same company's agent: if you shared your work email in an earlier conversation, we attach it to your new conversation so you are not asked for it again and the answers stay relevant to your company. The identifier is never shared between different companies' agents and is not used for advertising or analytics. Clicking "Restart" only starts a new conversation; your acceptance and the identifier stay until you delete this site's data in your browser. The legal basis is Art. 6(1)(b) GDPR for providing the conversation you requested and Art. 6(1)(f) GDPR for keeping it consistent across your visits; storing the identifier on your device is strictly necessary for the service you asked for (section 25(2) no. 2 TDDDG).

9. Sharing And Subprocessors

We may share personal data with:

  • clients, when you interact with a Namarix deployment operated for that client; the workspace operator's partners, such as sponsors, guests and clients, and the operator's own team, who may receive company-level topic summaries (a company name or work domain and a summary of topics asked about, never your name, email address or the words of your question), as described in section 5;
  • service providers and subprocessors that host, store, secure, analyze, process, automate, communicate, and deliver the service;
  • integration partners authorized by us or our clients, such as CRM, webhook, messaging, enrichment, email, or workflow tools;
  • licensees of aggregated, company-level market insights (cloud provider partner teams, software vendors and partner-tooling vendors), and podcasts with an Audience Partnership, which receive the weekly company-level report described in section 4 and may pass a company name and a summarised topic on to their sponsors, guests and listeners for companies with at least 50 employees. None of them receives your name, email address, contact details or the words of your question;
  • professional advisers, such as lawyers, accountants, auditors, insurers, and security consultants;
  • authorities, courts, regulators, or other parties where required by law or necessary to protect rights, safety, security, or contractual interests;
  • buyers, successors, or transaction participants in connection with a merger, financing, restructuring, asset sale, or future GmbH transition.

Our current sub-processors, by category, are: our primary AI model provider, used for answers, embeddings, classification, summaries and transcription (OpenAI); a fallback provider that steps in only when a request to the primary provider fails, for example because of an outage, a usage limit or exhausted credit, and then performs the same tasks except transcription with the same models and the same content (Microsoft Azure OpenAI Service; stored data is kept in the EU (Sweden); processing may also take place outside the EU); an emergency AI model fallback used only when the primary provider is unavailable (Google Gemini); reranking of retrieved passages (Cohere); assignment of your questions to topic categories and automated quality checks of AI-generated text, for example whether an answer is supported by the retrieved passages (TypeSafe AI, USA; the provider receives the text of your question, individual statements of an answer and the passages retrieved for them); checking whether an email domain can receive email (DNS services of Cloudflare and, as a fallback, Google); public web search, used for the company look-up described above and for the website-scoped answer to a business question (Serper); public web page fetch, used to read the published content of a public company page during that same company look-up (Firecrawl); database, storage and administrator authentication (Supabase, EU); object storage and content delivery (Cloudflare, EU jurisdiction); application hosting (Vercel, EU region); background media transcription (Hetzner Online GmbH, Germany); error monitoring and operational alerting (Sentry, EU region, with email addresses masked); internal operational alerts and demo-request alerts (Slack); email (Google Workspace); sending a workspace's newsletter and its confirmation email (SMTP.com, J2 Global Canada, Inc.); marketing website hosting (Webflow); and, where a client enables them, that client's CRM and notification destinations (HubSpot, Google Sheets, Slack, or a webhook the client configures). The always-current list, with the data each one receives and the transfer safeguard that applies, is published in our Sub-processor Register at https://namarix.ai/subprocessors.

When a request to our primary AI provider fails, for example because of an outage, a usage limit or exhausted credit, we may send the same request to Microsoft's Azure OpenAI Service instead. That request contains the same content that would otherwise have gone to OpenAI: for an answer, your question, the recent messages of your conversation, the source passages we retrieved to answer it, and our instructions to the model. Data that Microsoft stores for this service is kept in the European Union (Sweden), but for this service we use a setting that lets Microsoft process requests in any region where the model runs, so a request may be processed outside the EU. According to Microsoft, content sent to this service is not used to train its or OpenAI's models and is not shared with other customers. Microsoft may keep content that its automated safety systems flag, for abuse monitoring and review by authorised Microsoft staff, for a limited period.

Not every vendor receives every user's data. Which destinations receive your conversation depends on what the client operating that agent has configured. The summary described in section 5a goes to the CRM of the operator of the agent, where enabled.

10. International Transfers

We may process and transfer personal data in Germany, the European Union, the United States, and other countries where we or our service providers operate.

Some subprocessors are based in, or may process data from, the United States. Where we transfer personal data from the EEA, UK, or Switzerland to countries without an adequacy decision, we rely on appropriate safeguards such as standard contractual clauses, the EU-U.S. Data Privacy Framework where applicable, data processing agreements, and supplementary measures where required. Where a provider does not yet have a signed data processing agreement and we do not claim a transfer safeguard for it, we say so directly rather than let this paragraph be read as covering it. Three exceptions, which we state plainly rather than leave to be assumed. Our web-search provider (Serper) and our web-page-fetch provider (Firecrawl) are both based in the United States, our reranking provider (Cohere) processes data in the United States and Canada, and we do not yet have a signed data processing agreement with any of the three. We are putting those agreements in place. Until they are signed we do not claim standard contractual clauses for these three providers, and we keep what is sent to each of them narrow: for the company look-up, the domain part of an email address and the address of a public web page; for the website-scoped answer, at most 160 characters of the question you asked; for reranking, your question, candidate passages from the workspace library and, where known, a short company-context snippet of at most 160 characters. Our Sub-processor Register carries the current status for each provider.

Where Microsoft processes data outside the EEA in providing Azure OpenAI Service, the transfer is covered by the standard contractual clauses in Microsoft's data processing addendum.

Our sending service SMTP.com (J2 Global Canada, Inc.) is established in Canada and belongs to the U.S.-based Ziff Davis group. By its own account its email marketing business participates in the EU-U.S. Data Privacy Framework and otherwise relies on standard contractual clauses for transfers from the EEA. We have not yet signed a data processing agreement with SMTP.com, and until we do we do not claim standard contractual clauses for it. Its current status is in our Sub-processor Register.

11. Retention

We retain personal data only for as long as reasonably necessary for the purposes described in this Privacy Policy, including to provide the service, maintain security, comply with law, resolve disputes, enforce agreements, and support legitimate business needs.

Unless a different period is required by law, contract, or client instruction:

  • Account and workspace data: for the life of the account or client relationship, plus up to 24 months.
  • AI agent conversation history and session data: up to 36 months from the last interaction.
  • Click and action events under section 5a: 90 days, then daily counts only, with no link to a conversation or a person.
  • Source-of-visit data, topic and company assignment: like the conversation they belong to.
  • Signal records under section 5a: deleted together with the conversation they belong to.
  • Summary in the CRM: according to the retention periods of the CRM record, for Namarix up to 36 months from the last interaction.
  • Prospect email and campaign engagement data: up to 36 months from last meaningful engagement.
  • CRM, sales, and marketing records: up to 36 months from last engagement for inactive prospects, and for the active business relationship plus up to 24 months for customers.
  • Demo requests (request data, company profile, verdict, sample workspace): 6 months after the request or the last exchange. IP hash for rate limiting: 7 days.
  • Aggregated, de-identified analytics, product learnings and company-level market insights: indefinitely, provided they no longer identify an individual and are not used to re-identify them.
  • Our copies of the weekly reports to podcasts with an Audience Partnership: up to 36 months from the week they cover.
  • Workspace newsletter (subscription): while you are subscribed.
  • Workspace newsletter (unconfirmed request): 30 days.
  • Workspace newsletter (proof of sign-up, confirmation and unsubscribe): until 3 years after the end of the calendar year in which the subscription ends.
  • Workspace newsletter (send log): 12 months.
  • Copies of sent emails at SMTP.com: up to one year, per SMTP.com.
  • Marketing suppression records: as long as needed to honor opt-outs.
  • Security logs (IP address and request metadata): 180 days, unless a specific record is needed longer for security, abuse prevention, legal claims, or compliance.
  • Contract, invoice, tax, bookkeeping, and compliance records: for the statutory period required by applicable law, which may be up to 10 years for certain business and tax records in Germany.

Our primary AI model provider, OpenAI, is used under its API terms, under which content submitted to it is not used to train its models. According to Microsoft, the same applies to our failover provider, Microsoft's Azure OpenAI Service (see section 9). We have not yet obtained the same written confirmation from our other AI and search providers, so we do not make that promise for them; the current status for each is published in our Sub-processor Register at https://namarix.ai/subprocessors. Our own retention periods apply to copies stored in Namarix, client, CRM and backup systems. The full schedule is published as our Data Retention Schedule at https://namarix.ai/data-retention.

12. Security

We use technical and organizational measures designed to protect personal data against unauthorized access, loss, misuse, alteration, and disclosure. These measures may include access controls, encryption in transit, credential and secret management, logging, backups, provider security controls, and internal access restrictions.

No system is perfectly secure, and we cannot guarantee absolute security.

13. Your Rights

Depending on your location, you may have rights to:

  • request access to personal data;
  • request correction;
  • request deletion;
  • request restriction of processing;
  • object to processing based on legitimate interests;
  • request portability;
  • withdraw consent where processing is based on consent;
  • lodge a complaint with a data protection authority.

Namarix is established in Lüneburg, Lower Saxony. The competent supervisory authority is Die Landesbeauftragte für den Datenschutz Niedersachsen (the Lower Saxony Data Protection Commissioner): https://lfd.niedersachsen.de

To exercise your rights, contact mark@namarix.ai. If your data was processed on behalf of one of our clients or podcast partners, we may forward your request to that client or partner or help it respond.

Newsletter. You can withdraw your newsletter consent at any time, with effect for the future: through the unsubscribe link in every issue, through your email app's unsubscribe button, or by writing to mark@namarix.ai. You can also object at any time, without giving reasons, to our choosing the recommendations from your questions (Art. 21(2) GDPR). Because we only offer the newsletter with those picks, we then end your subscription.

Your right to object to company-level market insights and reports

You can object at any time to our use of your interactions for company-level market insights by writing to mark@namarix.ai. We will exclude your contributions from all future insights. An organisation can also ask us to exclude it entirely, and we will then leave it out of every future insight. Company-level topic summaries: you can also object at any time to your company being included in the company-level topic summaries that a workspace operator receives and may share, by writing to mark@namarix.ai. Tell us your work domain and, if you know it, the workspace you used. From then on we exclude that company's data from the summaries we prepare for sharing, and we tell the workspace operator, who must stop sharing entries about that company within 30 days. Summaries already shared before your objection cannot be recalled by us, but the workspace operator must not share them again. Reports to podcasts with an Audience Partnership: you can also object at any time to your contributions being used in these reports, and an organisation can ask us to leave it out entirely, by writing to mark@namarix.ai with your work domain. We then leave you, or that organisation, out of every future report, and tell each podcast that received an earlier report mentioning that company, which must not share entries about it again.

Your right to object to measurement

Objection to measurement and summary. Write to mark@namarix.ai. If you gave your email address, please state it: we then delete the events, the source, topic and company assignment of your conversations and the summary in the CRM, and add your address to a suppression list so that no new summary is created. If you gave no address, we cannot attribute events to a person (Art. 11 GDPR); in that case tell us the session identifier from the "kc_thread_" entry in your browser's local storage so that we can find the data concerned. We confirm receipt and reply without undue delay, at the latest within one month. If it concerns a client's agent, we forward your objection to the client or support the client in replying. No message to you is based on the summary.

If an AI-generated statement concerns you personally and you believe it is incorrect, you can also report this to mark@namarix.ai; we will review the report and work towards a correction.

14. U.S. Privacy Rights

If you are located in a U.S. state with applicable privacy rights, you may have rights to know, access, correct, delete, port, or opt out of certain uses or disclosures of personal information.

Workspaces operated by our clients and podcast partners. Where a client or a podcast partner operates a workspace, we process your information only as its service provider and processor. We never sell or share that information, and we never use it for the company-level insights or the reports to podcasts described below. Unless the client has turned it off, we may include it in the deidentified, aggregated weekly usage figures described in section 2, which cannot reasonably be linked to you or to any company, to improve the quality of our services; we maintain them only in deidentified form, do not attempt to reidentify them, and never use that information to profile you or market to you.

Company-level insights from workspaces we operate ourselves. From interactions with workspaces we operate ourselves, we produce company-level insights that are aggregated and deidentified, and we license them to businesses such as cloud provider partner teams, software vendors and partner-tooling vendors. We maintain and use this information only in deidentified form, we do not attempt to reidentify it, and we contractually require every recipient not to reidentify it. We also give podcasts with an Audience Partnership a weekly company-level report from these workspaces, as described in section 4. We maintain and use it only in deidentified form, we do not attempt to reidentify it, and our terms with each podcast prohibit it from reidentifying anyone, combining the report with other information to do so, and selling or licensing it. We do not sell or share personal information, as those terms are defined in the California Consumer Privacy Act, and we do not use personal information for targeted advertising. You can still opt out at any time: write to mark@namarix.ai with the subject "Opt out of company-level insights" and we will exclude your contributions from all future insights and reports. An organisation can also ask us to exclude it entirely.

If we ever use advertising or tracking technologies that count as a "sale", "sharing" or "targeted advertising" under applicable law, we will provide the required notices and opt-out methods first.

15. Children's Privacy

Our services are intended for business users and are not directed to children. We do not knowingly collect personal data from children under 16.

16. Demo Requests

This section applies when you use the "Get a demo" form on our website. Everything else in this policy still applies.

What you send us. Your full name, your work email address and, if you choose to add them, the RSS feed of your podcast or blog. The form is sent to our application server at app.namarix.ai. We also keep a one-way hash of your IP address. We use it only to limit repeated submissions. Because a hash can still be linked back to a device, we treat it as personal data and delete it after 7 days.

What we do with it.

  • We read your request and reply to you.
  • Company look-up. From your email address we use only the domain part (never your name or the part before the @). We run a public web search on that domain (serper.dev), fetch the company's public homepage (Firecrawl) and use an AI model (OpenAI) to summarise the result into a short company profile: company name, industry, headquarters city, approximate employee range, product line and typical customers. We do not query personal contact databases and we do not look you up personally. Free and disposable email domains are not looked up. We also count the public episodes or posts at the links you gave us.
  • Qualification. A fixed set of rules compares the company facts and the counts with our criteria and returns "qualified" or "not qualified". We use this only to decide in which order we handle requests. It does not decide whether you hear from us, a person reviews every request, and no one is refused or treated differently by an automated decision.
  • Internal alert. Your name, email address, the company facts and your links are posted to a private channel in our Slack workspace.
  • Sample workspace. Only after we have reviewed your request and decided to go ahead, we may build a sample workspace from the public content at the links you sent. We download the public audio of episodes, transcribe it with OpenAI, index blog posts and generate page text such as titles, descriptions and suggested questions. The sample is reachable at an address on namarix.ai that is not listed by search engines and works for whoever has the link. We email you the link from mark@namarix.ai (sent through Google Workspace). If you send no links, we build nothing. Questions asked in your sample are visible to us. They are not sold, not shared with anyone else and not used for the company-level market insights described in section 4.

People named in your content. Episodes and posts can contain the names, voices and job titles of other people, such as guests, hosts and authors. We take these details from the content itself, use them only to build the sample and do not build profiles of these people. If you are named in someone else's content shown in a sample, write to mark@namarix.ai and we remove it.

Why we may do this. For answering your request, arranging the demo and building the sample you asked to see, our legal basis is Article 6(1)(b) GDPR (steps at your request before a possible contract). For the company look-up, the qualification, the internal alert, the IP hash and the handling of people named in your content, it is Article 6(1)(f) GDPR (our legitimate interest in handling business requests efficiently, putting the most relevant first and protecting the form from abuse). We balanced this against your interests: we use little data, the look-up describes your organisation from public sources and not you personally, the results are seen only by us, and you can object at any time. You can request a summary of that assessment at mark@namarix.ai. Giving your name and email is necessary for us to handle the request. The links are optional.

Who receives it. Only our service providers: OpenAI (summaries, transcription, embeddings and page text; if a request to OpenAI fails, Microsoft's Azure OpenAI Service performs the same tasks except transcription as a fallback, see section 9), serper.dev and Firecrawl (domain only), Supabase (database, EU), Vercel (hosting, EU region), Cloudflare (storage and delivery), Hetzner Online GmbH (transcription, Germany), Slack (internal alert) and Google Workspace (email). Some are based in the United States, see section 10. We do not sell this data, do not give it to our clients and do not use it to train AI models.

About section 5. The statement in section 5 that Namarix sends you messages only if you subscribe to our newsletter or join our community applies to the AI agent. Here you asked us to get in touch, so we reply to your request. We do not add you to a mailing list or an automated sequence. Any further message follows section 7.

How long we keep it. Your request data, the company profile, the verdict and the Slack alert are deleted 6 months after your request or after our last exchange with you, whichever is later. The sample workspace, with its audio, transcripts and generated text, is deleted on the same date, or earlier if you ask. The IP hash is deleted after 7 days. If you become a client, the data moves into the customer record and section 11 applies.

Your rights. Section 13 applies. You can object at any time to the company look-up and the qualification by writing to mark@namarix.ai. We then handle your request without them and delete the result. You can also ask us to delete your request and your sample workspace at any time.

17. Changes

We may update this Privacy Policy from time to time. The updated version will be posted on our website with a new "Last updated" date. If changes are material, we may provide additional notice where required.

18. Contact

Young Hoon Lim Operating under the Namarix brand Hasenburger Ring 51d 21335 Lüneburg Germany mark@namarix.ai https://namarix.ai