Terms of Use (EULA)

Version: 2026-10-16-v9 · Last updated: 2 October 2026 (Automation agents - new section 13a: Auto / Ask / Never permission modes, the Approvals inbox, tasks on a schedule, via webhook and on demand, Credits charged per run, history, log and retention periods, WhatsApp numbers registered in the Provider's account, business users and data processing; new definitions in section 2; messages sent to third parties on your behalf - section 3.4(e) and new section 8.7; clarifications in sections 3.1, 10.3, 13.1-13.4, 14 and 15.4, including the current availability of Connectors and Meta features; a disabled video provider removed from section 3.5) · Effective date: 16 October 2026 (until 15 October 2026 inclusive, version 2026-09-27-v8 applies) · Source version: Polish - madd.im/regulamin.html

Provider / merchant: Marcin Kisielinski, sole trader operating under the business name MADD Marcin Kisielinski (brands MADD / DomaAI), registered office and address for service: ul. Kajki 10-12, 10-547 Olsztyn, Poland, VAT ID (NIP) 7422297084, REGON 545106682, entered in the Polish CEIDG register (hereinafter the „Provider”, „we”, „MADD”). Contact: [email protected]

1. General provisions

1.1. These Terms of Use set out the rules for using the DomaAI mobile application (iOS) and its web version available to users at domaai.pl (the same Service is also available at the existing address madd.im) (together: the „Application” or the „Service”).

1.2. The Service is provided by MADD. These Terms constitute the terms of service for the provision of services by electronic means within the meaning of the Polish Act of 18 July 2002 on the provision of services by electronic means.

1.3. Use of the Application requires acceptance of these Terms and of the Privacy Policy. Starting to use the Service constitutes the conclusion of an agreement for the provision of services by electronic means on the terms set out herein.

1.4. Language. The contract may be concluded in Polish or in another language in which the Application makes these Terms available. The Polish version is the source version and the others are translations of it; where you are a consumer, the version in the language in which the contract was concluded with you prevails, and a translation may not worsen your position.

1.5. Who these Terms apply to. They apply to: (a) consumers (Article 22¹ of the Polish Civil Code); (b) sole traders with consumer rights - natural persons entering into a contract directly connected with their business where the contract is not of a professional nature for them (Article 7aa of the Polish Consumer Rights Act, Article 385⁵ of the Civil Code); the rules on unfair contract terms, on conformity of digital content with the contract and on the right of withdrawal apply to them; and (c) other business users, subject to sections 10.4 and 10.7.

1.6. The Service is directed at users in Poland and the rest of the European Economic Area. It is not directed at sanctioned jurisdictions or at persons listed under restrictive measures.

2. Definitions

• Application / Service - the DomaAI application described in section 1.1 together with its backend.
• User - a natural person using the Application on the basis of a registered Account.
• Consumer - a User who is a consumer within the meaning of section 1.5(a).
• Sole trader with consumer rights - a User within the meaning of section 1.5(b).
• Account - the User's individual account created in the Application.
• User Content - prompts, files, data and other material the User enters into the Application.
• AI Content - text, images, video, documents, audio and other output generated by artificial intelligence models in response to the User's prompts.
• Credits - an internal, non-monetary unit of account used to access features of the Application (see section 6 and the Payments, subscriptions and refunds document).
• Subscription - a paid plan supplied continuously for a Billing Period and renewed automatically until cancelled.
• Billing Period - one month (30 days) or 12 months, depending on the Subscription variant chosen.
• Agent - the autonomous multi-agent system and related execution tools, including code execution and browser use (see section 5).
• Social module - features that make content available to other people: publishing to the social feed, group chats and sharing a conversation by link.
• Durable medium - a medium that allows information to be stored unchanged, in particular e-mail or a downloadable file.
• Apple IAP - the In-App Purchase mechanism operated by Apple Inc.
• Stripe - the online payment provider (Stripe Payments Europe, Ltd. / Stripe, Inc.) handling purchases in the web version of the Service.
• Connector - a link between the Application and the User's account in an external service (including Gmail, Google Calendar, GitHub, WhatsApp Business and - once made available - Google Ads, Shopify and Meta; section 13.1), established through the provider's official authorisation mechanism (OAuth or - for WhatsApp Business and Telegram Bots - an access token issued by the provider), within which the Agent acts on the User's instruction (section 13).
• Ad account - the User's advertising account in the Meta system (Facebook/Instagram) on which the Agent - on the User's instruction - creates and manages campaigns (section 14).
• Hosted website - a website or application built by the Agent on the User's instruction and published by the Provider at an address in the domaai.pl domain (section 15).
• Automation agent - a set configured by the User: name, role, instructions, AI model, Credit limit per run and permission modes, which carries out Tasks on the Provider's servers on the User's behalf on a schedule, when a webhook is called or on demand - also when the Application is closed (section 13a).
• Task - an instruction assigned to an Automation agent together with a schedule (daily, weekly or every given number of hours) or a webhook address and the method of delivering the result.
• Run - a single execution of a Task by an Automation agent (including on demand, a trial run or a continuation after a decision on an Approval), charged in Credits and recorded in the history.
• Permission mode - the User's setting for each capability of an Automation agent: Auto (the agent performs the action on its own), Ask (the agent prepares the action and waits for the User's decision in the Approvals inbox) or Never (the agent has no access to the capability).
• Approval - a request recorded in the Application by which an Automation agent asks to perform one specific action (with its full content, recipients, dates and amounts), which the User approves or rejects.
• Bot - an AI assistant configured by the User that conducts conversations with third parties on the User's behalf in WhatsApp or Telegram (section 13a).

3. What the Application is and the nature of AI Content

3.1. DomaAI is an artificial-intelligence assistant with a multi-agent architecture. The Application supports, among other things: text and voice conversations, image and video generation and editing, document creation, web search, code execution, browser use by the Agent, calendar and reminder management, projects, long-term memory and personalised answers, a private (incognito) mode that is not saved to history, group chats, sharing a conversation by link, publishing content in the social module, rating and reporting answers, support delivered by an AI assistant with escalation to a human, notes and transcripts from online meetings (at the User's invitation), monitoring of public-procurement notices with e-mail alerts, Automation agents carrying out Tasks on a schedule, via webhook or on demand, and WhatsApp and Telegram Bots (section 13a), a workflow builder, a website-building panel (WebStudio/Code), hosting of websites built by the Agent at an address in the domaai.pl domain (section 15), Connectors to external services - currently Gmail, Google Calendar, GitHub and WhatsApp Business (section 13) and, once made available, running ads on the User's Meta Ad account and publishing to the User's Facebook Page or Instagram (sections 13.1, 14 and 15.4) - and Code mode - execution of Agent actions on the User's device after pairing with the CLI. The feature set depends on the plan and may change (sections 3.7 and 6.6).

3.2. AI Content is generated automatically and is provided for information and assistance only. It may be inaccurate, incomplete, out of date or wrong (so-called „hallucinations”). You must verify AI Content yourself before relying on it.

3.3. AI Content does not constitute legal, medical, financial, tax, investment or any other professional advice. For matters requiring specialist knowledge, consult an appropriate professional.

3.4. In accordance with Article 50 of the AI Act (Regulation (EU) 2024/1689): (a) we inform you that you are interacting with an artificial intelligence system; (b) content generated or modified by AI is marked as artificially generated or manipulated in a machine-readable format, as far as technically feasible for the format concerned - images in file metadata, video files in container tags, Hosted websites in the meta tags of the HTML document; images and video generated or modified by AI additionally carry a visible watermark („AI · DomaAI”), and Hosted websites a visible notice that the content was prepared with AI; speech in voice mode is streamed without a resulting file, so the information that you are interacting with AI is given in the Application's interface. The metadata marking is not a cryptographic signature and can be removed with a metadata-editing tool (see the Privacy Policy, section 9.4); (c) a User who makes public generated content depicting existing persons, places or events in a way that could falsely appear authentic (a deep fake) has their own disclosure obligation as a deployer of an AI system; (d) you must not remove, obscure or alter AI content markings, including provenance metadata and notices that content was generated by AI, nor instruct third parties to do so (see section 8.5); (e) e-mails, WhatsApp messages and other communications that the Agent - in a conversation or as an Automation agent - sends to third parties on your behalf leave from your accounts as your messages and are currently not automatically labelled as prepared with an AI system (neither in technical message headers nor by a note in the content). Where the law requires informing the recipient that content was generated by AI or that they are communicating with an AI system, that obligation rests with you as the sender - you can fulfil it, for example, through an instruction to the agent or a note in the content of the Approval you approve. Bots tell the person they are chatting with that they are talking to an AI system in their first reply (section 13a.6).

3.5. The Application uses AI models supplied by third parties (including OpenAI, Anthropic, Google, Mistral AI, xAI, Cartesia, Tavily). The full list is in the Sub-processor list. DomaAI is not a product or service of those companies and is not affiliated with them. How content sent to models is handled - including confirmation that User Content is not used to train models - is described in AI data and model training.

3.6. AI supervision. The authority competent for AI systems in Poland is the Commission for the Development and Safety of Artificial Intelligence (KRiBSI), established by the Polish Act of 3 July 2026 on artificial intelligence systems (Journal of Laws 2026, item 1003). If you believe that use of the Service infringes the AI Act you may - independently of a complaint to us (section 12) - lodge a complaint with that authority (Article 85 of Regulation (EU) 2024/1689).

3.7. Experimental features. Some features may be released in a test form (beta, early access). They are marked in the Application, may be unstable and may be changed or withdrawn. Withdrawing a feature that is material to a paid plan is treated as a change to the scope of the Service and follows the procedure in section 4.5.

4. Account and requirements

4.1. Full use of the Application requires an Account created by:

• registering with an e-mail address and password (with e-mail verification by one-time code),
• signing in with „Sign in with Apple”, or
• signing in with a Google account (Sign in with Google).

4.2. Age requirement. The Application may be used only by persons aged 16 or over. On registration you confirm that you are at least 16 (Article 8 GDPR). Persons under 16 may use the Service only with the consent of a parent or legal guardian; the Service does not currently support such consent, so accounts for persons under 16 are not created.

4.3. Account security. You must provide true information, protect your login credentials and not share your Account with third parties. You are responsible for actions taken from your Account. Two-factor authentication (TOTP) is available in the Application and we recommend enabling it. Notify us at [email protected] immediately if you suspect unauthorised access.

4.4. Using the Application requires a device with internet access and, for the mobile version, a supported version of iOS; for the web version, a current version of a mainstream browser (Chrome, Safari, Firefox, Edge) with JavaScript and cookies enabled.

4.5. Changes to these Terms. We may amend these Terms for important reasons, namely: (a) changes in law or its interpretation, or decisions of authorities; (b) changes to the scope or manner of providing the Service, including adding or withdrawing features; (c) security reasons or the prevention of abuse; (d) changes on the side of third-party providers. We will give notice in the Application and by e-mail at least 14 days before the change takes effect. If you do not accept the change you may terminate the agreement with immediate effect (by deleting your Account or by a statement sent to [email protected]; a subscription must additionally be cancelled as described in section 11.4) up to the effective date; if you have an active paid subscription we will refund the proportional part of the fee for the unused period. Changes have no retroactive effect and do not affect rights acquired in the current, paid billing period.

4.6. Communication. We send statements concerning the agreement to the e-mail address linked to your Account and, additionally, as in-app messages; e-mail counts as a durable medium. You must keep your e-mail address up to date. Marketing messages are sent only on the basis of separate consent, which you may withdraw at any time.

4.7. One Account. You may hold one Account. Creating accounts to circumvent limits or blocks, reselling or sharing access to an Account, and offering the Service to third parties as your own service are prohibited, unless we agree otherwise under an Enterprise plan.

5. Agent, code and browser features

5.1. On your instruction, the Agent may autonomously, among other things: execute code (Python, shell commands) in an isolated environment, use a web browser, download files from the internet (up to 20 MB), search for and process information, and create files and documents.

5.2. Code and commands are executed in an isolated sandbox with no network access inside the sandbox, and the Agent's network access is filtered for security (protection against access to internal resources). Despite these precautions you use these features at your own risk and must not instruct the Agent to do anything unlawful or contrary to the Acceptable Use Policy.

5.3. Content downloaded or processed by the Agent from external sites and services comes from third parties; we are not responsible for its content, legality or accuracy.

5.4. Code mode (actions on your device). Once you deliberately pair the Application with the CLI, the Agent executes commands in your own environment, outside our sandbox. You decide which permissions and which working directory to grant, and you are responsible for the results of the operations performed, including changes to files. We recommend working on a copy of your data and using version control. You can unpair at any time in the Application settings.

6. Payments, plans and Credits

6.1. The Application is offered on a freemium basis. There is a free Free plan and paid subscription plans: Starter (base), Plus, Pro, Max, and an individually negotiated Enterprise plan. Use of features is billed in Credits granted with the plan and/or purchased in packs.

6.2. The payment method depends on the platform: in the mobile application (iOS) payments are made exclusively through Apple In-App Purchase (Apple is the merchant of record); in the web version (domaai.pl) payments are made through Stripe (secure Stripe Checkout page; the Provider is the merchant). In neither case do we store or process payment card data - it is handled by Apple or Stripe respectively.

6.3. Billing periods and renewals. A Subscription may be purchased on a monthly (30-day) or annual (12-month) basis. It renews automatically for another period of the same length until cancelled - in Apple ID settings on iOS, or in the Stripe customer portal on the web; cancel at least 24 hours before the end of the current period. On an annual subscription the Credit allowance is topped up every month (12 consecutive top-ups within the paid year) rather than made available once for the whole year; an unused monthly allowance does not carry over to the next month unless stated otherwise.

6.4. Prepaid plan. The web version also offers a 30-day prepaid plan (including payment by BLIK) which does not renew automatically. We send an e-mail reminder before the period ends, after which the plan reverts to Free.

6.5. Detailed rules on prices, packs, Credit accrual, subscriptions, cancellation and refunds are set out in a separate document, Payments, subscriptions and refunds, which forms an integral part of these Terms.

6.6. We may change prices, limits and Credit allocation rules for important reasons and on the terms described in Payments, subscriptions and refunds (section 2.3), with at least 30 days' notice for renewable subscriptions. Changes do not affect billing periods already paid for and active; you may cancel before renewal at no cost.

6.7. Purchases by minors. Only an adult may purchase a subscription, a Credit pack or a booster. A User aged 16 or over but under 18 may make purchases only with the consent of their legal representative (parent or guardian); by making a purchase you confirm that you are an adult or that you have such consent.

6.8. Nature of Credits. Credits are an internal unit of account of the Service. They are not electronic money or a payment instrument, are not exchangeable for cash and are not transferable between Accounts. This does not limit your right to a refund of the price under section 7 and the Payments, subscriptions and refunds document.

6.9. Accounting documents. For purchases in the web version we issue an invoice; business users provide their invoicing details (including VAT ID) before purchase. For purchases on iOS, receipts are issued and made available by Apple in your Apple ID purchase history.

7. Right of withdrawal (digital content)

7.1. Credit packs (digital content). Credits are delivered at your request immediately after purchase. A consumer loses the right to withdraw from a contract for the supply of digital content if they gave prior express consent to performance beginning before the withdrawal period expires and acknowledged the loss of that right, and the Provider gave them confirmation on a durable medium (Article 38(1)(13) of the Polish Consumer Rights Act).

7.2. Subscriptions (a digital service supplied continuously). A consumer may withdraw from a subscription contract - monthly or annual - within 14 days of its conclusion without giving a reason. If, at the consumer's express request, performance began before that period expired, on withdrawal the consumer pays an amount proportionate to what was supplied up to the moment of withdrawal (Article 35 of the Consumer Rights Act).

7.3. How to withdraw. An unequivocal statement sent to [email protected] is sufficient; you may use the model form attached to Payments, subscriptions and refunds (section 7). We refund promptly and no later than 14 days after receiving your statement, using the same payment method. Purchases made on iOS are settled by Apple - refund requests go to Apple (reportaproblem.apple.com) and we will assist you.

7.4. Sections 7.1-7.3 do not limit rights arising from statutory conformity/warranty rules for digital content, or the refund rules described in Payments, subscriptions and refunds (Apple for iOS, Stripe/the Provider for the web version). The right of withdrawal also applies to sole traders with consumer rights (section 1.5(b)).

8. Rules of use (acceptable use)

8.1. You undertake to use the Application in accordance with the law, good practice and the Acceptable Use Policy, which forms an integral part of these Terms.

8.2. The following are prohibited in particular: generating unlawful content (including child sexual abuse material, threats of violence, weapon-making instructions), infringing third-party rights, attempting to circumvent security measures, decompilation and reverse engineering, automated mass querying (scraping) of the Service, disrupting its operation, and using the Service to train competing AI models.

8.3. Prompts submitted to the image and video generation features are moderated (local filter plus an external moderation system). Content that breaches the rules may be blocked before the operation is performed.

8.4. Circumventing model safeguards („jailbreaking”), persistently repeating a request after a refusal, and using the Service to generate content that sexualises minors are also prohibited; such behaviour is recorded and may lead to the measures in section 8.6.

8.5. AI content markings. Removing, obscuring or altering markings and metadata indicating that content was generated or modified by AI - and distributing tools for doing so - is prohibited (section 3.4(d)).

8.6. Where these Terms are breached we apply measures proportionate to the seriousness of the breach: a warning, restriction or suspension of access to selected features or to the Account, removal of the infringing content and, in the case of a serious or repeated breach, termination of the agreement (section 11). Where a person frequently submits manifestly illegal content or manifestly unfounded notices we may suspend the service after a prior warning (Article 23 of the Digital Services Act). Every such decision may be appealed (section 11.5).

8.7. Messages to third parties. The Agent (in a conversation or as an Automation agent) and Bots may send messages only to people with whom you have a basis for contact (e.g. a reply to an enquiry, handling an order, an existing business relationship, the recipient's prior consent). Using the Agent or a Bot for any of the following is prohibited: mass or serial sending of unsolicited commercial communications (Article 398 of the Polish Electronic Communications Law), contacting people who have not consented where consent is required, circumventing platform rules (in particular the WhatsApp Business Policy and Meta's rules), impersonating others and sending content contrary to section 8.2. A breach results in the measures in section 8.6, including pausing the Automation agent's Tasks, disabling the Bot and disconnecting the Connector; for a number registered in the Provider's WhatsApp Business account (section 13a.9) - also deregistering the number from that account.

9. Rights to content (intellectual property)

9.1. Rights to the Application. The Application, its code, interface, trade marks, the name „DomaAI” and its documentation are owned by the Provider or its licensors and are protected by law. These Terms do not transfer those rights to you; you receive only a non-exclusive, non-transferable licence to use the Application as intended, on devices you own or control.

9.2. User Content (input). You retain all rights to the content you enter into the Application (prompts, files, data). You grant the Provider a non-exclusive licence limited to what is necessary to provide the Service (processing your prompt, passing it to model providers in order to generate a response, storing history). The licence ends when the content or the Account is deleted, subject to backups deleted on the cycle described in the Privacy Policy. User Content is not used to train AI models (see AI data and model training).

9.3. AI Content (output). To the extent permitted by law the Provider claims no rights to AI Content generated for you, and you may use it, including commercially, at your own risk. You acknowledge that:

• the same or similar AI Content may be generated for other users,
• copyright protection of fully machine-generated content may be limited or excluded,
• AI Content may inadvertently infringe third-party rights; verification before use is your responsibility.

9.4. Publishing and sharing AI Content in the social module, in group chats and by sharing a conversation link is governed by the Publishing and Sharing Policy. Sharing a conversation by link means anyone who knows the link can open it, until you revoke the share.

9.5. Feedback and reports. If you rate an answer, report a bug or send a suggestion, we may use that feedback for diagnostics and to improve the Service, without compensation and without time limit. A report includes the rated answer together with the necessary context; do not include data you do not wish to disclose.

10. Liability and availability

10.1. We make every effort to keep the Service working correctly, but provide it „as is” and, to the extent permitted by law, do not guarantee uninterrupted or error-free availability or the fitness of AI Content for a particular purpose. This does not affect the statutory obligation to supply digital content and a digital service in conformity with the contract towards consumers and sole traders with consumer rights (section 10.5).

10.2. The Service depends on third-party providers (AI models, infrastructure, Apple). Interruptions or changes on their side may affect how the Application works. We announce planned maintenance in the Application where possible.

10.3. To the extent permitted by law we are not liable for damage arising from: (a) decisions made on the basis of AI Content; (b) use of the Agent features (code, browser), including Code mode on your device, and actions performed by an Automation agent in Auto mode or after you approved an Approval, including the content of messages sent to third parties on your behalf; (c) the operation of third-party providers, including providers of services linked through Connectors (e.g. Meta's decisions to reject an ad, restrict an Ad account or charge advertising spend); (d) your breach of these Terms; (e) actions performed by the Agent in connected services on your instruction or with your consent (sections 13-15 and 13a). These limitations do not exclude or limit liability that cannot be excluded under mandatory law, in particular towards consumers and sole traders with consumer rights (including for personal injury, intent and gross negligence).

10.4. To the extent that our liability cannot be excluded and you are not a consumer or a sole trader with consumer rights, it is limited to the fees you paid in the 12 months preceding the event and does not cover lost profits.

10.5. Conformity with the contract (consumers and sole traders with consumer rights). We have a statutory obligation to supply digital content and a digital service in conformity with the contract. In the event of non-conformity a consumer may demand that conformity be brought about, and if that is impossible or excessively difficult, or has not happened within a reasonable time, may declare a price reduction or withdraw from the contract (Articles 43h-43n of the Consumer Rights Act). Complaints: section 12.

10.6. Force majeure. Neither party is liable for failure to perform caused by events beyond its reasonable control (including backbone network and provider infrastructure failures, cyber-attack, acts of public authority, natural disaster). This does not release us from refunding fees for the period in which the Service was unavailable, nor does it limit consumer rights.

10.7. Indemnity (applies only to Users who are neither consumers nor sole traders with consumer rights). You will indemnify the Provider against third-party claims and reasonable defence costs arising from your unlawful use of the Service or from content you publish.

11. Duration and termination

11.1. The agreement is concluded for an indefinite period. You may stop using the Service and delete your Account in the Application settings at any time, without giving a reason and free of charge.

11.2. Deleting your Account starts a 30-day recovery window (soft delete) during which you can restore the Account. After that, data is permanently deleted (see the Privacy Policy). Where a subscription is active, permanent deletion takes place no earlier than after the end of the paid period (Article 17(3)(b) GDPR). You can download a copy of your data in the Application before deleting your Account.

11.3. We may terminate the agreement or suspend an Account for important reasons, in particular in the event of a serious or repeated breach of these Terms, action harmful to the Service or to other users, or where required by law. Where possible and lawful we will give you prior notice. If we terminate for reasons not attributable to you, we refund the proportional part of the fee for the unused period.

11.4. Cancelling a subscription is not the same as deleting an Account and is done - depending on where it was purchased - in Apple ID settings (iOS) or in the Stripe customer portal (web) (see Payments, subscriptions and refunds).

11.5. Statement of reasons and appeal (Digital Services Act). If we restrict the visibility of content, remove it, suspend an Account or terminate the agreement, we notify you with a clear and specific statement of reasons: what the decision concerns, its legal or contractual basis (a provision of law or a clause of these Terms or the Acceptable Use Policy), whether automated means were used in the assessment, and how to appeal (Article 17 of Regulation (EU) 2022/2065). You may appeal free of charge within 6 months of being notified, at [email protected]; appeals are decided by a human, not by an automated system, within 14 days (Article 20 DSA). You also have the right to choose a certified out-of-court dispute settlement body (Article 21 DSA) and to go to court. How to report illegal content is described in Reporting illegal content.

11.6. Assignment. We may transfer our rights and obligations under the agreement to another entity (for example on a change of legal form or a transfer of the business) only if this does not worsen your position. We will give 14 days' notice; you may then terminate the agreement with immediate effect and receive a proportional refund. You may transfer your rights under the agreement only with our consent.

12. Complaints

12.1. Complaints about the operation of the Service can be sent to [email protected], stating a description of the problem and the e-mail address linked to the Account.

12.2. We respond to complaints within 14 days of receipt. If we do not respond within that period, the complaint is deemed accepted. The response is provided on a durable medium - to your e-mail address.

12.3. Out-of-court dispute resolution. A consumer may use out-of-court methods of handling complaints and pursuing claims, including: (a) permanent arbitration consumer courts at the regional trade inspectorates; (b) mediation conducted by the regional trade inspector; (c) free assistance from a district (municipal) consumer ombudsman or consumer organisations. Details: polubowne.uokik.gov.pl. Use of these procedures is voluntary; the Provider does not undertake in advance to participate in ADR proceedings - it states its position in its reply to the complaint.

12.4. Consumers in other EU Member States may use the free assistance of the European Consumer Centre for their country. The EU ODR platform was shut down on 20 July 2025 and no longer accepts complaints.

13. Connectors and integrations (Gmail, Google Calendar, GitHub, WhatsApp, Meta, Google Ads, Shopify, MCP)

13.1. What Connectors are. At your request the Application connects to your accounts in external services, as well as to Apple calendar and reminders (EventKit) and to MCP servers you specify. As at the publication of this version of the Terms, the following can be connected: Gmail, Google Calendar, GitHub and WhatsApp Business. The Google Ads and Shopify Connectors are provided for in the Service but are currently not available for connection. The Meta Connector (Ad account, Facebook Page, Instagram) is visible in the Application (labelled “Requires Meta approval”), but Meta has not granted the Application the permissions needed for ads and Pages, so an attempt to connect it is currently refused. The provisions on those Connectors (including sections 14 and 15.4) apply from the moment connecting them becomes possible. You can see the Connectors and their connection status in the Application (Account → Connectors). Google's verification of the Application for the Gmail and Google Calendar Connectors is in progress - until it is completed Google may show a warning when connecting that the app has not been verified. The connection is made through the provider's official authorisation mechanism (OAuth 2.0), and for WhatsApp Business and Telegram Bots - through an access token issued by the provider (in the Meta dashboard or via @BotFather) and pasted by you, or by confirming the number with a code sent by Meta (section 13a.9); tokens are stored in encrypted form (section 13.6). The Application never asks for the password to an external service. You choose the scope of access yourself and may revoke it at any time in the Application (Account → Connectors) or in the provider's settings; disconnecting deletes the token on our side and revokes it with the provider where the provider allows it (for Google - when the last Google Connector is disconnected, because revocation at Google covers all of your Google Connectors).

13.2. The Agent acts on your instruction. In a connected service the Agent does only what follows from your instruction - given in a conversation or recorded in an Automation agent's Task and in the permission modes you granted it (section 13a) - within the granted scope of access. You are responsible for the content of your instructions and for being entitled to use the connected account (e.g. a work account) in this way. Every action of the Agent on a Connector is recorded in a log visible in the Application (retention: section 13a.7 and Privacy Policy, section 7).

13.3. Consent gate for irreversible actions. Actions whose effects cannot be undone - sending an e-mail, publishing a post, inviting guests to an event, changing or publishing a store product, creating an issue in a repository, enabling an advertising campaign - are performed by the Agent only after it has shown you exactly what it intends to do (recipient, content, date, amount) and you have explicitly agreed - in the conversation or, for an Automation agent, in the Approvals inbox (Ask mode, section 13a.3). The consent mechanism is built into the Application and cannot be bypassed by the model. The exception is Auto mode granted by you to a given capability of an Automation agent (section 13a.2): it is a prior, standing instruction to perform actions of that kind without separate consent for each of them; for capabilities marked in the Application as requiring confirmation (including sending e-mails and WhatsApp messages, changing calendar events and publishing) it requires a separate confirmation in the Application, and it is not available for enabling ad budget spend (section 14.2) or for changing products in an online store. Reading data (e.g. searching a mailbox, listing events, campaigns or orders) requires no separate consent, as it follows directly from the instruction.

13.4. Data from connected services. Data retrieved from an external service is processed ad hoc, solely to carry out the specific instruction, and is not copied into a separate database nor indexed; excerpts may remain in the history of the conversation, which you can delete. For an Automation agent, excerpts may also remain in the result of a Run and in the content of an Approval - for the periods set out in section 13a.7 and in the Privacy Policy, section 7. Data from Google services is used in accordance with the Google API Services User Data Policy, including the Limited Use requirements - it is not used for advertising, sold or used to train AI models. Details: Privacy Policy, section 2.5, and AI data and model training, section 6.

13.5. Providers' terms and liability. Use of a Connector is also subject to the terms and policies of the external service provider (including Google, GitHub, Shopify, Meta), and that provider remains an independent controller of your data in its service. We are not responsible for the operation, availability, limits or decisions of external services (e.g. restriction of API access, suspension of an account with the provider) nor for the consequences of instructions you give. Connectors may be released in stages (e.g. in a provider's test mode, for a limited number of users) until the provider has completed its review of the Application - we inform you of this in the Application.

13.6. Credential security. Access tokens for connected services are stored only in encrypted form, on an encrypted volume; they are not made available to AI models or third parties. The configuration of integrations (including MCP servers) is excluded from data export.

13.7. Point of contact (DSA). The Provider's single point of contact - both for Member State authorities, the European Commission and the European Board for Digital Services (Article 11 of the Digital Services Act) and for recipients of the service (Article 12 DSA) - is the e-mail address [email protected]. Communication may be in Polish or English.

13a. Automation agents and Bots

13a.1. What an Automation agent is. You may create an Automation agent (section 2) and assign Tasks to it. The Automation agent carries out Tasks on the Provider's servers - on a schedule, when a webhook is called (an address to which you or your system sends data, e.g. a form submission) or on your request - also when the Application is closed, using the Connectors you have connected (section 13) and the Agent features (section 5). The result of a Run is delivered in the way you choose - to your Account e-mail address or to a selected conversation in the Application, optionally with a push notification. You can perform a trial Run in which the agent describes what it would do without performing actions that have effects outside the Application and without creating Approvals. An Automation agent makes no decisions about you within the meaning of Article 22 GDPR; it carries out your instructions. The provisions of section 13a on liability, messages to third parties and data processing apply accordingly to Bots.

13a.2. Permission modes. For each capability of an Automation agent you set a permission mode:

• Auto - the agent performs the action on its own, without asking; the action is recorded in the Run history and in the log,
• Ask - the agent prepares the action and records it as an Approval in the Approvals inbox; the action is performed only after you approve it,
• Never - the agent has no access to the capability.

Default modes depend on the role chosen for the agent and follow the principle of least privilege: capabilities not needed for the role are set to Never, reading data usually to Auto, and actions with effects outside the Application (sending a message, changing a calendar event, publishing, changing a campaign) to Ask; code execution is set to Auto only in roles that need it, and enabling ad budget spend, changing products in an online store and creating further Tasks are set to Never by default. Setting Auto mode for a capability marked in the Application as requiring confirmation requires a separate confirmation in which the Application describes the effect of that setting; the confirmation is recorded. Auto mode cannot be set for enabling ad budget spend (section 14.2) or for changing products in an online store. For the browser and code execution only Auto and Never are available, because these actions cannot be held until you decide. The number of actions with effects outside the Application that the agent may perform in Auto mode in a single Run is limited (currently to 5); further such actions in that Run are not performed automatically but go to the Approvals inbox and wait for your decision (section 13a.3).

13a.3. Approvals. An Approval contains a description of the action prepared by the agent: recipients, content, attachments, dates and amounts. Before approving you may correct the fields indicated in the Application (e.g. the body or subject of a message); you approve what you see. One Approval concerns one specific action and is performed once. In the iOS app, approving a high-risk action may require identity confirmation on the device (Face ID, Touch ID or device passcode). An Approval on which you have not decided within 7 days expires and the action is not performed; we may send one reminder. The number of new Approvals from a single Run and the number of Approvals awaiting a decision are limited (currently 5 and 30 respectively) - once the limit is reached the agent creates no further Approvals and says so in the result. Once all Approvals from a Run have been decided, the agent may complete the Task in a continuation Run, charged in accordance with section 13a.5. An Approval approved by a person who has gained access to your device or Account is attributed to you under section 4.3.

13a.4. Responsibility for instructions and messages. The actions of an Automation agent - in Auto mode and after an Approval is approved - and the replies of Bots are your actions performed on your instruction, in your name and from your accounts. You are the sender of messages sent by the agent and the Bot and the publisher of published content; you are responsible for the content of Tasks and instructions, for the legal basis for contacting the recipient (section 8.7), for informing recipients about the processing of their data and about the use of AI where the law requires it (section 3.4(e)), and for compliance with providers' terms (section 13.5), including the WhatsApp Business Policy and Meta's rules. Where you act as a business, you entrust the Provider with processing the data of your customers and contacts on the terms of section 13a.10.

13a.5. Credits for Runs. Every Run - including trial Runs, on-demand Runs and continuations - is charged in Credits: a start cost is charged upfront when the Run begins, and the cost of the model's work and of the tools used is charged according to the price list in the Application - in total up to the Credit limit per Run you set for the agent (within the limits shown in the Application; 60 Credits by default). The limit cannot be lower than the start cost of the given Run plus 10 Credits - if the limit you set is lower (in particular for Tasks run on the agent's virtual machine, whose start cost is higher), the higher value applies to that Run. The amount charged does not exceed the limit applicable to the given Run; once it is used up the agent stops and describes what it managed to do. Before a Task is saved the Application shows an estimate of the cost of one Run and a monthly forecast - these are indicative values; the actual cost depends on the work performed and is shown in the Run history. An active Task consumes Credits on every Run until you switch it off; you may at any time switch off a Task, pause all Tasks, change the limit or delete the agent. If your Credit balance is insufficient for the start cost, the Run is skipped and not charged. The start cost of a Run that ends in an error or is interrupted (e.g. by a server restart) is returned to the Credit pool it was taken from; a Run that fails for transient reasons may be retried automatically. After three consecutive failed Runs the Task is paused automatically and you are notified. We may apply plan-dependent limits on the number of agents, active Tasks and Run frequency, as well as daily limits on Runs and spend (Payments, subscriptions and refunds); the limits that apply to you are shown in the Application. A scheduled Task runs not more often than hourly. A webhook Task runs on every accepted event (each webhook call is a separate, charged Run) - within the daily event limit for the Task (100 per day by default) and the limits of your plan.

13a.6. AI information towards third parties. WhatsApp and Telegram Bots tell the person they are chatting with that they are talking to an AI system in the first reply of a conversation. Messages an Automation agent sends to third parties are currently not automatically labelled - your obligations in this respect are set out in section 3.4(e).

13a.7. History, log and retention. We record: the Run history (result, cost charged, status) - 180 days from the Run (the last 20 Runs of each Task are kept longer, until the Task or the Account is deleted); input data sent via a webhook (e.g. form content) - 24 hours from the Run, after which the history only shows that it has been deleted; the agent's action log (tool, time, result, a shortened preview of arguments with credentials masked; the content of messages sent only as its length) - 90 days; the Connector activity log (section 13.2: type of action, object identifier and data describing the action - e.g. recipients and subject of an e-mail, title and guests of an event, title of a document or issue; without the content of messages sent) - until the Account is deleted, and entries created in a Run - also when that Run is deleted; Approvals - the full content of the action (arguments) 30 days from the decision or expiry, the remaining Approval data (status, time, tool, description and a preview of the action without message content, e.g. recipients and subject) 90 days from the creation of the Approval. You may delete a Task's history or a single Run (together with the related Approvals and log entries) earlier in the Application; deleting a Task deletes its history, and deleting the Account deletes all of this data (section 11.2). Details: Privacy Policy, sections 2.6 and 7.

13a.8. Stopping and security. You may at any time switch off a Task, pause all Tasks, deactivate or delete an agent, disable a Bot and disconnect a Connector. Disconnecting a Connector prevents the agent from further using that service; an Approval that needs a disconnected service stays in the Approvals inbox and, if approved, is not performed but returns to waiting with a request to reconnect the service - after reconnecting you may approve it again. An Approval expires after the period in section 13a.3. An Approval also cannot be performed once the capability is set to Never or the agent is deactivated or deleted - until that period ends it stays in the Approvals inbox, and an attempt to approve it ends it as expired. We may pause Runs (globally, for a User or for a Task) for security reasons, in the event of a provider outage, abuse or exceeded limits, informing you in the Application; this does not constitute a lack of conformity of the Service with the contract if the pause is necessary and proportionate. An Automation agent operates within the safeguards described in section 5.2 (sandbox, traffic filtering, protection against data exfiltration); despite them, actions in Auto mode are performed without your ongoing involvement, which you take into account when granting modes.

13a.9. WhatsApp number registered in the Provider's account. You may connect your own phone number to a Bot without creating a Meta account: we then register that number in our WhatsApp Business account (WhatsApp Cloud API) as a hosted number. You confirm your right to the number with a code sent by Meta; the number remains your number, and its display name is subject to Meta's approval. Use of a hosted number is charged in Credits for 30-day periods according to the price list in the Application; the fee for the next period is charged with the first message after the paid period ends, until you disconnect the number. Messages from a hosted number weigh on the Provider's WhatsApp Business account, so you comply with the WhatsApp Business Policy and Meta's messaging rules (including the 24-hour reply window and the ban on unsolicited messages), and we may apply lower daily sending limits to hosted numbers than to own numbers, suspend sending or deregister the number from our account in the event of a breach of section 8.7 or if Meta lowers the number's quality rating. The number of hosted numbers is limited. You may disconnect the number at any time - it is then deregistered from our account. You may also make a Bot publicly available on the DomaAI WhatsApp number: a person who starts a conversation from the „Start @name” link you share talks to your Bot until they write „menu” or the Bot is disabled or deleted; the Bot's replies are charged in your Credits, and these conversations are conversations of your Bot within the meaning of sections 13a.4 and 13a.10.

13a.10. Business users - data processing on your behalf. A User who uses an Automation agent or a Bot in the course of business or professional activity (including sole traders with consumer rights) and instructs the Provider to process personal data of third parties (customers, contractors, Bot contacts, people submitting a form connected to a webhook) is the controller of that data, and the Provider is a processor acting on the User's instructions (Article 28 GDPR). By accepting these Terms the parties conclude a data processing agreement to that extent with the following content:

• subject matter and duration - carrying out the Tasks and conversations configured by the User, for as long as the User uses these features;
• nature and purpose - storage, organisation, consultation, transfer to the AI model provider chosen by the User in order to generate content, and sending communications - solely to provide the Service to the User, never for the Provider's own purposes (including not for advertising, profiling or model training);
• categories of data subjects and data - the User's customers, contractors, contacts and correspondents: identification and contact data, the content of messages, conversations, submissions and attachments; the User does not instruct the processing of special categories of data (Articles 9 and 10 GDPR);
• instructions - the User's documented instructions are the configuration of the agent, the Task, the Bot and the permission modes, and every approved Approval; the Provider processes the data only on such instructions, including with regard to transfers to a third country (transfer to the AI model provider chosen by the User for the agent or Bot, listed in the Sub-processor list, is made on the User's instruction, on the basis of the mechanisms in Chapter V GDPR), unless required to do so by Union or Member State law - in which case we inform the User of that legal requirement before processing, unless that law prohibits it; the Provider immediately informs the User if, in its opinion, an instruction infringes the GDPR or other data protection provisions;
• the Provider's obligations - confidentiality of authorised persons; the security measures described in the Privacy Policy, section 10 (Article 32 GDPR); use of the sub-processors listed in the Sub-processor list, on whom the Provider imposes data protection obligations no less stringent than in this section and for whose acts it is liable as for its own (Article 28(4) GDPR) - we give 14 days' notice of changes to the list and the User may object and stop using the feature; assistance with data subjects' rights (the User can view and delete the history, Approvals, Bot conversations and the logs in the Application) and - to the extent of the information available to the Provider - with the obligations under Articles 32-36 GDPR (security, breach notification, impact assessment, prior consultation); notifying the User of a personal data breach without undue delay and no later than 48 hours after becoming aware of it, together with the information referred to in Article 33(3) GDPR; after the User stops using the feature - at the User's choice - deleting the data or enabling the User to download it before deletion (data export in the Application, section 11.2), unless law requires further storage, and absent another decision of the User the data is deleted within the periods in section 13a.7 and the Privacy Policy, section 7; making available the information necessary to demonstrate compliance and allowing for and contributing to audits, including inspections, conducted by the User or an auditor mandated by the User - with at least 14 days' prior notice, not more than once a year (unless the audit follows a personal data breach or a request of a supervisory authority), without infringing the Provider's trade secrets or other users' data; in the first place by providing documentation and answers to audit questions;
• the User's obligations - ensuring a legal basis for the processing and for contact (section 8.7) and providing information to third parties (Articles 13 and 14 GDPR).

If the User has concluded a separate data processing agreement with the Provider (DPA template), that agreement prevails over this section to the extent it governs. This does not apply to the User's conversation with their own assistant on the DomaAI WhatsApp number - there the Provider remains the controller.

14. Meta ads on the User's Ad account

14.1. Principle and availability. The feature described in section 14 is available only once Meta grants the Application the required advertising permissions and you connect your Ad account. As at the publication of this version of the Terms Meta has not granted those permissions, so an Ad account cannot be connected (section 13.1). Once your Meta Ad account is connected the Agent may - on your instruction - create campaigns, ad sets and ads, change, pause, copy and delete them, estimate reach, select audiences and read results. Ads are created on your account and in your name: you are the advertiser and we provide the tool. Meta bills advertising costs directly to you according to the payment method set on your Ad account; we take no commission on ad spend, and use of the feature is billed in Credits like other Agent features.

14.2. Everything is created paused. Every campaign, ad set and ad created by the Agent is created in the paused state (PAUSED) and generates no spend. An ad is enabled (ACTIVE) only after your explicit consent - given in the conversation or, for an Automation agent, by approving an Approval in the Approvals inbox - once the name, daily budget, audience, content and preview of the ad have been presented (the consent gate in sections 13.3 and 13a.3), or by you directly - in the Application or in Meta Ads Manager. The „enable budget spend” capability cannot be set to Auto mode (section 13a.2).

14.3. Budget cap. You set a daily budget cap in the Application (by default PLN 100 per day per ad set). The Agent cannot set a daily budget above the cap nor raise the cap itself - only you can raise it. The cap applies to the budget set by the Agent; actual spend is charged by Meta under its own rules (including the possibility of exceeding the daily budget on individual days while keeping the weekly average), over which we have no control.

14.4. Advertiser obligations. You are responsible for the compliance of ads with the law (including unfair competition, consumer rights and regulated-product advertising rules) and with Meta's advertising policies, including the accuracy of the payer and beneficiary information required by the Digital Services Act (Article 26 DSA), for the rights to the graphics and texts used, and for labelling AI-generated content where the law or Meta's rules require it. Commissioning ads that are discriminatory, misleading, concern prohibited goods or services, or political and electoral ads without meeting legal requirements is prohibited.

14.5. Meta's decisions and liability. Meta may reject an ad, pause it, restrict the Ad account or change the terms of access to its interface; we have no influence over this and are not liable for such decisions or for campaign results (reach, clicks, conversions, cost). Ad content proposed by the Agent is AI Content within the meaning of section 3 - you verify it before enabling. The availability of the feature depends on the permissions Meta grants to the Application (section 14.1).

14.6. Log and revoking access. All actions of the Agent on the Ad account are recorded in a log visible in the Application. You may disconnect the Ad account at any time (section 13.1); disconnecting does not delete campaigns previously created on your account in Meta.

15. Hosted websites and publishing by the Agent

15.1. Website hosting. On your instruction the Agent may build a website or simple application and publish it at an address in the domaai.pl domain (a Hosted website). A Hosted website is your content: you decide on its publication, content and removal from the address, and we act as a hosting service provider within the meaning of the Digital Services Act. The rules of publishing, responsibility for content, reporting of illegal content and moderation - including statement of reasons and appeal - are set out in the Publishing and Sharing Policy.

15.2. AI marking of websites. Hosted websites, as content built by an AI system, receive on publication machine-readable meta tags in the HTML document and a visible notice that the content was prepared with AI (section 3.4). You do not remove these markings (section 8.5); you may add your own markings and information required by law (e.g. seller details, terms, privacy policy), for which you are responsible if the website serves your business.

15.3. Duration and limits of hosting. Website hosting is billed in Credits for the period indicated in the Application (e.g. 30 days) and depends on the plan; after the paid period the website ceases to be public and its files remain in the Application under the retention rules. We may limit the size, traffic or number of websites and take down a website that breaches these Terms, the Acceptable Use Policy or third-party rights, or that endangers security (e.g. phishing, malicious code), following the procedure in section 11.5.

15.4. Publishing to external services. Publishing a post to a Facebook Page, Instagram or another connected service is performed by the Agent only after your consent (section 13.3) - in the conversation or in the Approvals inbox - or in Auto mode granted by you to that capability after a separate confirmation (section 13a.2), in your name and to your account. Publishing to Facebook and Instagram is possible only once Meta grants the Application the required permissions - as at the publication of this version of the Terms they are not available (section 13.1). You are the publisher of such content, responsible for it towards third parties and the service provider and for the labelling of AI-generated content required by law (section 3.4(c)). Images and video from the Application are published with the „AI · DomaAI” watermark (section 3.4(b)).

16. App Store (Apple) provisions

16.1. Section 16 applies only to the Application downloaded from the App Store and constitutes the end-user licence agreement (EULA) required by Apple. In the event of a conflict with the other sections of these Terms, section 16 prevails to that extent.

16.2. Parties. These Terms are concluded between you and the Provider only, and not with Apple. The Provider alone is responsible for the Application and its content.

16.3. Scope of licence. You are granted a non-transferable licence to use the Application on any Apple-branded products that you own or control, as permitted by the Apple Media Services Terms and Conditions.

16.4. Maintenance and support. The Provider is solely responsible for maintenance and support services for the Application. Apple has no obligation whatsoever in that respect.

16.5. Warranty. The Provider is solely responsible for the Application's conformity with the contract. In the event of non-conformity you may notify Apple, and Apple will refund the purchase price of the Application (if any); to the maximum extent permitted by law Apple has no other warranty obligation whatsoever, and any further claims are directed to the Provider.

16.6. Product claims. The Provider, not Apple, is responsible for addressing your claims or those of third parties relating to the Application or your possession and use of it, including product liability claims, claims that the Application fails to conform to any applicable legal or regulatory requirement, and claims arising under consumer protection law.

16.7. Intellectual property rights. In the event of a third-party claim that the Application infringes intellectual property rights, the Provider alone is responsible for the investigation, defence, settlement and discharge of that claim.

16.8. Legal compliance. You represent that you are not located in a country subject to a US Government embargo or designated by the US Government as a „terrorist supporting” country, and that you are not listed on any US Government list of prohibited or restricted parties. You undertake to comply with applicable third-party terms of agreement when using the Application (for example your wireless data service agreement).

16.9. Contact details. MADD Marcin Kisielinski, ul. Kajki 10-12, 10-547 Olsztyn, Poland, [email protected] - address any questions, complaints or claims concerning the Application to that address.

16.10. Third-party beneficiary. Apple Inc. and its subsidiaries are third-party beneficiaries of this section 16 and, upon your acceptance of these Terms, have the right to enforce it against you as a third-party beneficiary.

17. Final provisions

17.1. Governing law. Matters not covered by these Terms are governed by Polish law, in particular the Civil Code, the Consumer Rights Act, the Act on the provision of services by electronic means, the GDPR, the Digital Services Act and the AI Act. The choice of Polish law does not deprive a consumer of the protection afforded by the mandatory provisions of the country of their habitual residence (Article 6 of the Rome I Regulation).

17.2. Jurisdiction. Disputes with a consumer are heard by the court of general jurisdiction; a consumer habitually resident in another EU Member State may bring proceedings before the courts of their place of domicile, and we may sue them only before the courts of that State (Articles 17-19 of the Brussels I bis Regulation). Disputes with Users who are neither consumers nor sole traders with consumer rights are heard by the court having jurisdiction over the Provider's seat.

17.3. These Terms do not limit consumer rights arising from mandatory provisions of law. Provisions contrary to such rules are not binding on a consumer.

17.4. If any provision of these Terms proves invalid or ineffective, the remaining provisions remain in force and the relevant provisions of law apply in place of the invalid one.

17.5. Our failure to exercise a right under these Terms is not a waiver of that right.

17.6. Accessibility. The Provider is a micro-enterprise, to which - as regards services - the Polish Act of 26 April 2024 on accessibility requirements for certain products and services (implementing Directive (EU) 2019/882) does not apply. Regardless, we work to keep the Application and these documents readable and accessible; please report any barriers to [email protected].

17.7. The current version of these Terms is available in the Application and at domaai.pl (and at the existing address madd.im). Each version is numbered in the format YYYY-MM-DD-vN; earlier versions are available on request at [email protected].

The following form integral parts of these Terms: Privacy Policy, Payments, subscriptions and refunds, Acceptable Use Policy, Publishing and Sharing Policy, AI data and model training and Reporting illegal content.