Privacy notice

1. The controller

• Controller / operator: Hegedős-Törzsök Jusztina Anett e.v.
• Registered office: 1016 Budapest, Aladár utca 14. As.1.
• Tax number: 57445892-1-41
• Sole proprietor registration number: 11699722
• Email: info@tendermatch.hu
• Telephone: +36 70 946 1064
• Website: tendermatch.hu

Under GDPR Article 37 we are currently not required to designate a data protection officer (DPO). We answer data protection questions at the email address above.

2. Applicable legislation

• Regulation (EU) 2016/679 — General Data Protection Regulation (GDPR)
• Act CXII of 2011 — the right of informational self-determination (Infotv.)
• Act CVIII of 2001 — electronic commerce services
• Act XLVIII of 2008 — commercial advertising activities
• Act C of 2000 — on accounting

3. What we use your data for

• Creating and maintaining your account — registration as an applicant, a grant writer or a project manager.
• Providing the service (matching) — based on your company data, we show you which grants you are eligible for, and connect you with experts who suit you.
• Communication — notifications about offer requests, offers and your account.
• Grant Alerts — an email about new calls matching the search you have saved.
• Invoicing and legal obligations — subscriptions, credit purchases, issuing invoices, bookkeeping.
• Web analytics — cookieless, aggregated statistics on how many people come to the site, where from, and what they use on it (details in section 10).
• Company directory and quick eligibility check — if you enter a tax number, we retrieve the company's public data (name, registered office, legal form, validity of the tax number) from the Online Invoice system of the National Tax and Customs Administration (NAV), and save it in our own company directory. We do not save who looked it up, nor the headcount or turnover given for the check.
• Assistant — in the chat that opens from the bottom right corner of the page, we search for grants for you and answer questions about using the platform. The answers are written by a language model from the platform's own data; to do this, it receives your messages (section 8). The masked text of unanswered questions may also be sent to a language model so that it can prepare suggestions for extending the answers (section 8). At your request, the same assistant answers your question from the text of a call, or writes a working draft of a grant application (AI draft): for this we use a language model (section 8).
• Bug reports and suggestions — if you report a bug or send a suggestion with the Feedback button, we save it linked to your account so that we can investigate it and, for an accepted one, add credits to your account.
• UX survey — if you fill in the short survey, your answers tell us what is useful on the site and what is missing. We choose the questions based on what we already know from your account (your role, and which features you have used), so we do not ask about these separately. We add credits to your account for filling it in. We show the reminder no more than three times.

4. On what legal basis

• Performance of a contract (GDPR Article 6(1)(b)) — operating the account, the matching and the handling of offers.
• Legal obligation (Article 6(1)(c)) — invoicing and accounting obligations.
• Consent (Article 6(1)(a)) — newsletters and marketing communications.
• Legitimate interest (Article 6(1)(f)) — system security, prevention of abuse, handling of any legal disputes, the cookieless web analytics (section 10), and filtering out undeliverable email addresses. Keeping the company directory is also our legitimate interest: storing the public company data of the tax authority makes the eligibility check faster and more accurate. The assistant's answers to your questions are based on our legitimate interest when you are not signed in (answering visitors' questions about the service), and on the performance of a contract when you are signed in (Article 6(1)(b)). Keeping the masked text of unanswered questions for 90 days is also our legitimate interest: it tells us what the assistant cannot answer yet and what we need to improve; so is preparing the suggestions with a language model. Handling feedback is also our legitimate interest: fixing the bugs reported and developing the service. The screenshot may also show whatever was on your screen at the time of the report; we mask password and tax number fields and, where a connection has been established, the other party's contact details before sending, and you can delete the image before sending. The UX survey is likewise our legitimate interest: filling it in is voluntary, and we choose what to ask about solely from our own data.

Grant Alerts are part of the service, so they relate to the performance of the contract; you can switch them off at any time without the other features of your account changing.

5. What data we process

• Identification and contact: name, email address, telephone number. If the applicant accepts an offer, the applicant and the chosen expert can see each other's name, email address and telephone number (if provided), so that they can contact each other.
• Company data: company name, tax number, TEÁOR code, region, headcount, turnover, equity, operating profit, scope of activities.
• Company directory: the company name, registered office and legal form belonging to the tax number looked up, and the validity of the tax number, from NAV's public data. For a sole proprietor this is personal data, since the name of the business is the proprietor's name.
• Assistant: the message you type, the page you are asking from, and, when you are signed in, a summary of your account (your role, how complete your company profile is and its business data, the number of your offer requests, your credit balance — not your name, email address or telephone number). We use these for the answer, on our own server; to write the answer, a language model also receives them together with the earlier messages of the conversation and, when you are signed in, together with your name (section 8). We store the question you ask about a call and the project description you give for a draft, together with the answer, linked to your account, so that you can look at them again later; if you delete your account, we also delete your questions and project descriptions. Experts see a draft only if you attach it to an offer request (we mask the contact details we recognise); when you delete your account, the attached draft is also removed from the request, unless you have already chosen an expert. We do not write the text of the conversation to a database (it stays in the server's memory for no more than 6 hours, section 8): your browser keeps it in the tab you are using, together with the signed state of the conversation (the company details given: region, legal form, headcount, turnover, year of foundation, TEÁOR; the topic and the programme, and the identifiers of the calls shown most recently; it contains no tax number). We record pseudonymised usage data (time, the type of page, whether you asked as a visitor or signed in, when signed in your role, the recognised intent and topic of the question, the kind of data given — not its value —, whether we managed to answer, response time). If we could not answer a question with certainty, we keep its text to improve the answers: up to 200 characters, with tax numbers, email addresses, telephone numbers, longer strings of digits, and parts that look like names and addresses automatically masked; if we recognise personal data in the question, we do not keep it, nor the answers given by tapping buttons. The masking is automated, so do not include personal data. The rows of one conversation are linked by a random conversation identifier; we do not record your IP address or your account with them. From the masked question text we keep, a language model may prepare suggestions each night for extending the answers (section 8); for this we check the masking again, and delete any text in which we now recognise personal data.
• Feedback: your description and its type (bug or suggestion); the address of the page you sent it from (without the values of the search parameters); a screenshot of the page as you saw it at the time (unless you deleted it); the element selected on the page; your last steps on the site in that browser tab before sending (pages opened, without the values of the search parameters, the labels of the buttons and links pressed, error messages, failed requests); the browser's type, language and window size. All of this linked to your account. What you typed into fields is not recorded among the steps.
• UX survey: your answers (willingness to recommend, 0–10, your rating of the features you used, why you did not try a feature, and what you wrote in your own words), and alongside them your situation at the time of filling it in, from your account data: your role, which features you have used and how many times (company profile, Grant Alerts, offer request, question about a call and draft writing, expert profile, sending offers), the age of your account; and when we showed you the reminder, and whether you asked to be reminded later. All of this linked to your account.
• Expert profile: field of expertise, grant types taken on, experience, references, fees.
• Usage data: offer requests, offers, ratings, saved searches, credit transactions.
• Billing data: billing name, address, tax number.
• Technical data: IP address, cookie identifiers, log data.
• Email data: if one of our emails cannot be delivered (it bounces), or the recipient marks it as spam, our email service provider (Resend) tells us, so that we do not send further emails to that address. The links in our emails contain campaign parameters (UTM), so the web analytics statistics show how many visits came from the emails — but not who clicked. We do not track the opening of emails or clicks on their links per person.
• Web analytics data: without cookies and without identifying the visitor — details in section 10.

6. Who we pass it on to (processors)

The providers below process the data on our instructions and for the purpose indicated here; they may not use it for their own purposes.

• Hetzner Online GmbH

  Industriestr. 25, 91710 Gunzenhausen, Germany

  Hosting and server services (running the platform and the database)

  Data transferred: All data stored on the platform, in the German data centre

• Clerk, Inc.Outside the EEA

  660 King St., San Francisco, CA 94107, USA

  User accounts, sign-in, password management, two-factor authentication

  Data transferred: Name, email address, profile picture, sign-in logs

• Resend, Inc.Outside the EEA

  2261 Market St., San Francisco, CA 94114, USA

  Delivery of transactional and notification emails

  Data transferred: Name, email address, the content of the message

• Anthropic PBCOutside the EEA

  548 Market St., San Francisco, CA 94104, USA

  Ranking the results, extracting the eligibility criteria from the text of the call, and answering a question asked about a call and writing a working draft of a grant application (AI draft); furthermore, writing the assistant's chat answers from the platform's own data (catalogue, eligibility rules, help texts); reading the intent and data of a message to the assistant that the rules do not understand with certainty; FAQ and synonym suggestions from unanswered questions, with human approval

  Data transferred: The text of the call for proposals; for search, the professional data of the company profile (activity, TEÁOR, size, region); for a question about a call and for a draft, the question or project description typed by the user, the questions they asked earlier about the same call and the business data of the company profile (company name – not for a sole proprietor –, legal form, region, TEÁOR, size, headcount, turnover, equity, year of foundation, activity), supplemented with what they said in the conversation (region, legal form, headcount, turnover, the topic and the amount requested). For these, we do not send the user's name, email address or contact details. For the assistant, also: the chat messages and the earlier messages of the conversation, the address and catalogue filter of the current page; for a signed-in user, their name, the business data of the company profile and the status of the account (offer requests, offers, credits), without email address and telephone number; the legal form and region looked up on the basis of a company tax number entered; the data of the calls; the chat message not understood with certainty, after the recognised personal data has been masked; the masked text of unanswered questions, up to 200 characters, without account or conversation identifiers

• Seresné Kurucz-Morvai Kitti e.v.

  3000 Hatvan, Határ utca 23.

  Bookkeeping, handling and retaining accounting documents

  Data transferred: Billing and accounting data

Integrated, but not active at present

These providers are connected to the system but are not currently switched on, so they do not process any data at present. As soon as we activate them, this notice will be updated.

• Cloudflare, Inc.Outside the EEANot yet active

  101 Townsend St., San Francisco, CA 94107, USA

  CDN, DDoS protection and TLS encryption in front of the website

  Data transferred: IP address, request metadata (no content is stored)

• Stripe Payments Europe, Ltd.Not yet active

  1 Grand Canal Street Lower, Dublin 2, Ireland

  Online card payments and subscription management

  Data transferred: Name, email address, payment data (we do not see card details)

• Barion Payment Zrt.Not yet active

  1117 Budapest, Irinyi József utca 4-20. II. em.

  Online card payments (alternative payment provider)

  Data transferred: Name, email address, payment data (we do not see card details)

• KBOSS.hu Kft. (Számlázz.hu)Not yet active

  1031 Budapest, Záhony utca 7.

  Invoicing, data reporting to the NAV Online Invoice system (Online Számla)

  Data transferred: Billing name, address, tax number, the items of the invoice

• OpenAI Ireland Ltd.Not yet active

  1st Floor, The Liffey Trust Centre, 117-126 Sheriff Street Upper, Dublin 1, Ireland

  Vectorising the text of calls for proposals for semantic search

  Data transferred: Only the text of public calls for proposals — no user or company data

7. Transfers outside the EEA

The legal framework for such transfers is Chapter V of the GDPR: the standard contractual clauses (SCCs) adopted by the European Commission are in force with the providers concerned, or — where the provider is certified — the adequacy decision on the EU–US Data Privacy Framework. The privacy notice of the provider concerned gives information on the exact mechanism, and we will also send it to you on request.

For our providers within the EEA (Hetzner Online GmbH and Seresné Kurucz-Morvai Kitti e.v.), there is no transfer outside the EEA.

8. Use of artificial intelligence

We also use a language model to search for grants, to extract the eligibility criteria, and to answer questions asked about a call and to write grant drafts (AI draft). It is important that you know the following about this:

• The data sent to the model is primarily the text of the public call for proposals. Your name, email address and contact details are never transferred.
• In a natural-language search, we send the professional data of the company profile (activity, TEÁOR, size, region) to rank the results, so that the results are relevant.
• When you use the question about a call and grant draft features (in the assistant, started with the buttons of a call), the question or project description you type and the questions you asked earlier about the same call are also sent to the model, together with the call's data. If you have a company profile, the business data of the company is also sent to the model: company name (not for a sole proprietor), legal form, region, TEÁOR, size, headcount, turnover, equity, year of foundation and activity; if you mentioned any of these in the conversation, that too (for example region, headcount, the topic and the amount requested). Do not include anything you would not like to share. We store the question or project description and the answer linked to your account, so that you can look at them again later.
• The chat answers of the assistant (in the bottom right corner) are written by a language model (Anthropic, as a processor) from the platform's catalogue, eligibility rules and descriptions checked by us. The model receives the message you type, the earlier messages of the conversation and which page you are on (with which catalogue filter); when you are signed in, also your name, the business data of your company profile and the status of your account (offer requests, offers, credits), but not your email address or telephone number. If you enter a company tax number, also the legal form and the region once they have been looked up. Our server keeps the text of the conversation in its memory for no more than 6 hours, so that it can continue at the next message; we do not write it to a database. If it does not know something for certain, it says so, and refers you to the official call or to customer service. Without signing in, you can ask 5 questions every 24 hours. Do not include personal data; a company tax number may be given: if you enter one, we retrieve the company's public data from the company directory or from NAV, and the company's public data is added to the directory (see above); we do not record that it was you who looked it up.
• We may also send the masked text of unanswered questions (up to 200 characters, without recognised personal data), grouped, to a language model (Anthropic, as a processor), so that it can prepare a suggestion for a new FAQ answer or synonym. We only send a question that was asked in at least two different conversations, and we do not send a conversation identifier, time or account data. A member of our team checks and approves the suggestion; until they approve it, nobody sees it. Here too, we delete the masked example sentences after 90 days.
• The model does not decide on eligibility. The ‘you are eligible / you are not eligible’ result is produced by deterministic rules, by comparing your company data with the conditions of the call — at most, the model reads the conditions out of the text of the call, and until a human has approved this, we show them marked ‘Preliminary’.
• We do not take any decision based solely on automated processing that would produce legal effects concerning you (GDPR Article 22). Next to every result we show why it is there, and the official call always overrides our indication.

9. How long we keep it

• Account data: for as long as the account exists, and after its deletion for no longer than the general civil-law limitation period (5 years).
• Invoicing and accounting data: for 8 years from issue (section 169 of Act C of 2000).
• Credit ledger: for the same period as the accounting retention period — credits represent a value expressed in money, so their settlement must remain traceable afterwards.
• Newsletter data: until you unsubscribe.
• Technical and log data: until the security purpose is achieved, or until an incident has been investigated.
• Company directory: for 3 years from the last query, after which we delete it automatically.
• Feedback: for 12 months from submission, after which we delete it automatically, together with the screenshot and the steps. Credits added to your account remain in the credit ledger for longer than this.
• UX survey: we keep the answers for 24 months from filling in, and the reminder data for 24 months from their last change, after which we delete them automatically. Credits added to your account remain in the credit ledger for longer than this.
• Assistant: we do not write the text of the chat to a database: our server keeps it in its memory for no more than 6 hours (section 8), and your browser in the tab you are using until you close it (the same applies to a project idea you described for the draft in the chat). We keep the questions you asked about calls and your drafts, together with the answers, until your account is deleted, and the copy of a draft attached to an offer request together with that offer request. We keep the masked text of unanswered questions, up to 200 characters, for 90 days, after which we delete it automatically. We keep the pseudonymised usage data for statistical purposes, not linked to your account. The masked example sentences saved with the answer suggestions are also deleted after 90 days; rejected suggestions after 12 months.
• Web analytics statistics: 25 months, after which we delete them automatically.

10. Cookies and web analytics

We use the cookies necessary for operation (session, sign-in, protection against abuse) on the legal basis relating to the provision of the service, or on the basis of legitimate interest; we do not ask for separate consent for these. We do not use analytics or marketing cookies. Should this change, we would place such cookies only with your prior, explicit consent, and the consent could be withdrawn at any time.

Sign-in cookies are managed by Clerk, with the HttpOnly, Secure and SameSite settings.

When you are signed in, your browser remembers your last few steps in the session storage (sessionStorage) of the tab you are using, so that we can attach them to a bug report. This does not leave your computer unless you send feedback, and it is deleted when the tab is closed. In the same place, it also remembers with a flag that we have already asked you about the UX survey reminder in this session, so that we do not ask again.

Without signing in, the assistant uses a cookie named tm_asst (HttpOnly, valid for 7 days): it contains a random browser identifier and a counter of how many free questions you have asked in the past 24 hours; we do not link it to your account. Your browser keeps the conversation and its signed state (the company details given, the topic and the calls shown most recently; without tax number), as well as the project idea you described for the draft, in the session storage (sessionStorage) of the tab you are using, and deletes them when the tab is closed.

Web analytics

To see how many people come to the site, where from, and what they use on it, we compile web analytics statistics. For this we use an open-source analytics tool called Umami, on our own server, in Hetzner's data centre in Germany; the tracking script loads from stats.tendermatch.hu. The data is not passed to third parties, and we do not use it for advertising.

• No cookies, and we do not store your IP address. We do not place any identifier on your device. From the IP address and the browser's user-agent string, the analytics server computes, one-way and with a secret key, an identifier that changes daily: within one day your visits count as one visitor, but visits on two different days cannot be linked. From the IP address it also determines the approximate location (country, county, town), but it does not store the IP address itself.
• What we record: the address (URL) and name of the page visited; where you came from (for another website, only its domain, e.g. facebook.com); the type of browser, operating system and device, the screen size and the language; the approximate location; and these actions: clicks on the sign-up buttons, submitting the choice of role (with the role chosen), sending an offer request, an offer request sent successfully (with the type of page it started from), the result of the quick eligibility check (match, no match or incomplete match; without the data entered), an alert saved successfully (with the type of save), a completed registration (with the role chosen), saving a search as an alert, sending a question about a call (for an answer written from the text of the call), requesting a grant draft (AI draft), saving the company profile, submitting the quick eligibility check (without the data entered), opening the assistant, sending a message to the assistant (without its text), reaching the assistant's question limit, clicks on the assistant's sign-up button, clicks on a call shown by the assistant, opening a question about a call or the writing of a draft in the assistant (the task and the place, without text) and attaching a grant draft to an offer request. From the URL we keep only the campaign parameters (UTM), and for pages within your account we also drop the unique identifiers. What you type (search, question, company data) is not recorded.
• Purpose: measuring the use of the service and the reach of our campaigns, and developing the service. We only look at aggregated statistics: we do not identify individual visitors, and we do not link the measurement to your account.
• Legal basis: our legitimate interest (GDPR Article 6(1)(f)). We have weighed it up: our interest — seeing what works on the site and which campaign brings visitors — does not disproportionately infringe your privacy, because the measurement takes place without cookies and without storing IP addresses, with an identifier that changes daily, exclusively on our own server.
• Retention: 25 months, after which we delete it automatically, so that the same period can still be compared across two years.
• Objection: if you switch on the ‘Do Not Track’ or the Global Privacy Control setting in your browser, the site sends no measurement data; the same applies if you block the tracking script. You can also send your objection to info@tendermatch.hu. As the measurement is not linked to your name or your account, we are generally unable to find your data among the data already recorded (GDPR Article 11).

11. Your rights

• Access: you can request information on what data we process about you, and request a copy of it.
• Rectification: correction of inaccurate or incomplete data.
• Erasure: where justified, you can request the erasure of your data, except data we are required by law to keep.
• Restriction: you can request that we only store your data and do not process it further.
• Objection: to processing based on legitimate interest. In the case of direct marketing, the objection takes effect immediately.
• Data portability: you can receive the data you have provided in a machine-readable format, or request that it be transmitted to another provider.

You can send your request to info@tendermatch.hu or by post to our registered office. We will reply on the merits without undue delay, and at the latest within 30 days; for a complex request this may be extended by a further 60 days, of which we will inform you within 30 days. Requests are fulfilled free of charge.

12. Remedies

If you feel that your rights have been infringed in connection with the processing of your data, you can lodge a complaint with the supervisory authority:

• Hungarian National Authority for Data Protection and Freedom of Information (Nemzeti Adatvédelmi és Információszabadság Hatóság, NAIH)
• Address: 1055 Budapest, Falk Miksa utca 9-11.
• Telephone: +36 (1) 391-1400
• Email: ugyfelszolgalat@naih.hu
• Website: naih.hu

You can also go to court — you can also bring the action before the regional court (törvényszék) of your place of residence or place of stay. We ask you to contact us first: we can resolve most issues more quickly directly.

13. Data security

• Data traffic is encrypted with TLS.
• We do not store passwords: sign-in is handled by Clerk, in line with industry standards.
• Access is role-based; users can only access their own data.
• Financial transactions (credits, payments) form an unalterable log that can be checked afterwards.
• Data minimisation: we only ask for what the service needs.

14. Changes to this notice

If the service or the range of processors we use changes, we will update this notice, and notify registered users of any material change by email.