Croche Terms of Service

1. Purpose

These Terms govern the use of Croche, provided by HIRE AI Inc. (operating as H1R AI; "we"). They set out the conditions and procedures of use and the rights, duties and responsibilities of us and of you.

2. Definitions

1. "Service" means everything we provide at croche.ai and related domains:
   • the Croche Runtime, SDK and CLI, distributed through npm and similar channels ("SDK");
   • Croche Studio, the web dashboard;
   • Croche Cloud: accounts, keys, usage limits and budgets, the keys and routing for AI model calls, usage records, config snapshot storage, and evaluation features such as virtual-user testing;
   • related docs and APIs.
2. "You" means the individual or entity that accepts these Terms and creates an account.
3. "Your App" means any app, website or software you build or operate with the Service.
4. "End Users" means the people who use Your App.
5. "Your Content" means the inputs that you or End Users send through the Service (prompts, conversations, images, settings, instructions and the like), and the outputs that AI models generate for them.
6. "Model Providers" means the third parties that operate AI models. This includes the routing provider we use, currently OpenRouter, Inc.
7. "Studio Plan" means the development usage limit, per period, that you can use through Studio and development keys while building.
8. "Runtime Credits" means the balance used for the AI usage of Your App once deployed.
9. "Keys" means project keys, developer keys, server keys, development keys, model-call keys and any other credentials we issue.

3. Effect; related documents

1. We post these Terms at croche.ai/terms and show them at sign-up. We highlight material terms (Sections 5, 9, 11 and 16) separately when you sign up.
2. The usage policy in Section 7 and the Croche Privacy Policy apply together with these Terms. A Data Processing Addendum (DPA) forms part of them only if one is signed.
3. Use of SDK source and binaries is governed by the licence distributed with the package. If these Terms conflict with that licence, the licence prevails for the SDK code itself, and these Terms prevail for Cloud and Studio.
4. Matters these Terms do not cover are governed by Korean law, including the Act on the Regulation of Terms and Conditions, and by commercial custom.

4. Accounts

1. You sign up with Google, GitHub or email. During the beta, you need a valid invite code that we distributed. Invite codes may not be transferred or sold.
2. You must be an individual aged 18 or older, or a legal entity. Persons under 18 may not use the Service. If you sign up on behalf of an entity, you warrant that you have the authority to accept these Terms for it. You may not sign up if the law of Korea or of your country prohibits your use, including export-control and sanctions restrictions.
3. You must give accurate information and keep it up to date.
4. You must keep your account and Keys secure. You are responsible for all usage under your account and Keys, including Keys embedded in Your App as distributed and usage through the unverified-install path we provide. Tell us at once and rotate the Key if it leaks or is misused. Per-key caps and similar safeguards limit exposure, but they do not mean we cover losses from a leak. This does not apply to losses caused by our intent or negligence.
5. We may refuse or later cancel a sign-up for false information, if you are under 18, for re-registration by someone we previously restricted, for an improperly obtained invite code, or for any other breach of these Terms.

5. Beta

1. The Service is in beta. It is provided "as is" and "as available", with no SLA, no availability commitment and no obligation to provide support.
2. Features, APIs, the protocol, data formats and pricing may change without backward compatibility, or be discontinued, before general availability. We give advance notice where we can. For security or emergency reasons we may give notice afterwards.
3. Beta data (config snapshots, logs, evaluation results and the like) may not be migrated, and may be deleted, when the Service is reworked. Keep your own copies of anything important.
4. We advise against running critical or safety-relevant production services on the beta. If you do, you bear that risk.
5. You may discuss the beta, its features and its performance publicly, except information we ask you in writing (email counts) to keep confidential.

6. Your responsibilities for Your App and End Users

1. You are solely responsible for Your App, its content and your relationship with End Users. If an End User's use of the Service through Your App breaches these Terms, that breach counts as yours.
2. You must:
   • publish your own terms and privacy policy for End Users, saying that we process their data as your processor and that their input is sent to third-party Model Providers, including abroad;
   • tell End Users that outputs are AI-generated where the law requires it, and comply with applicable AI laws, such as Korea's AI Basic Act;
   • give End Users a way to report inappropriate AI output, including where app-store policies require one;
   • obtain any End User consent or other legal basis needed, including for international transfers;
   • comply with the policies of your distribution channels, such as Google Play and the Apple App Store.
3. If Your App targets children or processes their personal information, you must meet the applicable legal requirements and tell us in advance.
4. You must evaluate outputs yourself for accuracy, fitness and safety. Outputs may be wrong, biased or inappropriate. Similar outputs may be produced for other users.

7. Usage policy

You must not do the following, and you must take reasonable steps to stop End Users from doing them:

1. break the law, or infringe others' rights (copyright, trademark, privacy, reputation, publicity and the like);
2. generate or distribute illegal or harmful content, such as child sexual abuse material, non-consensual intimate imagery, terrorism or weapons of mass destruction, malware, fraud or phishing, spam, or illegal gambling;
3. make decisions with major effects on a person's life, body, property or legal status (medical diagnosis, legal judgment, credit, hiring or insurance decisions and the like) from AI output alone, without human review;
4. make End Users believe the AI is a human, or impersonate real people without their consent;
5. send resident registration numbers or other unique identifiers, sensitive data or financial data to the Service without a legal basis;
6. share, resell or redistribute Keys or accounts to third parties without our consent. Use inside Your App, in the way the Service provides, is allowed;
7. circumvent or disable usage limits, budgets, key caps, device-integrity checks or access controls, including by creating multiple accounts to obtain credits or plans;
8. overload the Service or its infrastructure (including Model Providers), run vulnerability scans or penetration tests without approval, or try to access other users' data. If you find a security vulnerability, please report it to privacy@h1r.ai;
9. breach the usage policies of Model Providers. You must check the policies of the models you use.

8. Third-party Model Providers

1. Your App's model calls are sent with a Key we issue through a routing provider (currently OpenRouter, Inc.) to third-party Model Providers, which may be located in the US, China and other countries. During the beta, Model Providers based in China may be used; an option to exclude them is not yet available.
2. We configure routing to exclude providers that train on inputs. Providers may still retain data briefly under their own policies, for caching, abuse prevention or legal duties. We do not control or guarantee what providers actually do.
3. Model Providers decide model availability, performance, price, output quality and deprecation. We do not guarantee that any model stays available. We may change routing (the choice of provider) for quality, reliability or cost.
4. We are not liable for losses caused when a Model Provider restricts or suspends use. This does not apply where we acted intentionally or negligently.

9. Plans, credits and budgets

1. We do not take payment during the beta. Paid plans and credit top-ups will be added later; before they are, we will add fees, payment and refund terms to these Terms and give notice under Section 18.
2. The Studio Plan and Runtime Credits are separate. The Studio Plan is a usage limit for development; Runtime Credits pay for the AI usage of Your App once deployed. Usage is metered by our records and deducted from the limit or the credits, including usage from evaluation and virtual-user testing.
3. During the beta, the Studio Plan and Runtime Credits are promotional credits that we grant free of charge by promotion or code. Promotional credits have no cash value. They cannot be transferred, refunded or exchanged for cash, and they expire at the end date of the promotion, or 90 days after grant if no end date is stated. We may reclaim credits that were improperly obtained.
4. Budgets, usage limits and per-key caps are tools to help you manage usage. Because metering and aggregation lag, actual usage can somewhat exceed a limit you set. We may set or change per-user and per-project limits to protect the Service.
5. If a limit or credits were wrongly deducted through our fault, such as an outage, we restore them.

10. Our duties

1. We comply with the law and these Terms, and make reasonable efforts to provide the Service stably. The beta terms in Section 5 take precedence.
2. We apply reasonable security measures, including encryption in transit, least-privilege access, and hashed or encrypted storage of Keys.
3. We receive complaints and inquiries at privacy@h1r.ai, handle them within a reasonable time and tell you the result.

11. Data roles and privacy

1. For End User personal information and Your Content, you are the controller (the personal information controller under Korean law), and we are your processor. We process it only for the purposes you entrust to us: providing the Service, metering usage, security and troubleshooting. Article 26 of Korea's Personal Information Protection Act applies to this relationship.
2. Your End User privacy policy must name us as a processor and disclose the sub-processors and international transfers described in Section 8. You authorize our use of the sub-processors listed in the Croche Privacy Policy, including Cloudflare (hosting), OpenRouter (model routing) and Model Providers. We give notice of sub-processor changes by updating the Privacy Policy.
3. What we store. Retention periods follow the Croche Privacy Policy.
   • The Cloud stores call metadata for metering and debugging: identifiers, timestamps, model, token counts, cost and status.
   • The Runtime in Your App sends End User messages and AI responses directly to the routing provider; our Cloud does not receive or store their content.
   • Config snapshots and evaluation results you upload are kept until you delete them or close your account.
4. We do not use Your Content to train AI models, and we configure routing so that it is not sent to providers that train on inputs (Section 8(2)).
5. The internal LLM layer of the SDK (internal prompts, memory structures and the like) is not shown in the Service. You receive the inputs and outputs exchanged between Your App and the SDK.
6. We are the controller of your account data (email, name and sign-in provider identifiers) and of aggregated usage statistics used to improve and secure the Service (such as per-model latency, error rates and cost), under the Croche Privacy Policy.
7. DPA: we provide a DPA on request. Once signed, the DPA prevails over these Terms on personal-data processing.
8. If we receive a legal demand (from law enforcement, a court or the like), we notify you where the law allows and disclose only the minimum required.
9. If a personal-data breach occurs, we notify you within the period the law requires and help you meet your own notification duties.

12. Intellectual property

1. Your App, Your Content (including outputs), and the config snapshots and instructions you create belong to you (or their rights holders). You grant us a non-exclusive right to use them only as needed to provide the Service under these Terms. We do not warrant that outputs are legally protectable or unique to you.
2. The Service (SDK, Studio, Cloud, docs, the "Croche" trademark and logo) and its improvements belong to us. Your use of the SDK code is governed by the licence in Section 3(3). These Terms do not give you any right to use our trademarks, except for factual statements such as "Built with Croche".
3. We will not publish your name, logo or Your App as a customer reference without your consent in writing (email counts).

13. Feedback

If you give us suggestions, bug reports or opinions about the Service, we may use and incorporate them freely, without payment or obligation. We will not name you as the source without your consent. Feedback is provided "as is", and you make no warranty about it.

14. Suspension

1. We may suspend all or part of an account, project or Key if:
   • you breach these Terms or the usage policy;
   • there is a security or legal risk to the Service, other users, Model Providers or third parties;
   • invite codes, plans or credits were improperly obtained;
   • a Model Provider or a public authority requires it.
2. Before suspending, we tell you the reason and give you a chance to fix it. If there is an urgent security risk, illegal activity, or a risk of harm to third parties, we suspend first and notify you without delay.
3. Once the cause is resolved, we restore access within a reasonable time. You may object to a suspension at privacy@h1r.ai.

15. Termination and discontinuation

1. You may close your account at any time by writing to privacy@h1r.ai.
2. We may terminate if you commit a material breach, if you fail to cure a breach within 14 days of our notice, or if the law prevents us from continuing the Service.
3. We may end the beta or the whole Service, with 30 days' notice. Promotional credits and Studio Plans are not compensated or refunded.
4. On termination, Keys stop working and Your App's AI features may stop. For 30 days after termination you can ask us to export your data (config snapshots and usage history). After that we delete it, except records we must keep by law.
5. Sections 11(9), 12, 13, 16–18 and 20 survive termination.

16. Disclaimers and limitation of liability

1. Disclaimer. To the maximum extent the law allows, the Service and AI outputs are provided "as is". We make no express or implied warranty of merchantability, fitness for a particular purpose, uninterrupted or error-free operation, or accuracy of outputs.
2. Indirect damages. We are not liable for lost profits, lost data, business interruption or other indirect, special or consequential damages. This does not apply where we knew or could have known of such damage.
3. Cap. Our total liability relating to the Service is limited to the greater of (a) the amounts you actually paid us in the 12 months before the damage occurred and (b) KRW 100,000. Promotional credits and Studio Plans do not count as amounts paid.
4. Exceptions. Subsections (2) and (3) do not apply to damage caused by our intent or gross negligence, to harm to life or body, or to any liability that cannot be excluded by law.
5. We are excused to the extent that we cannot provide the Service because of events beyond our control, such as natural disasters, war, power failures, or outages at third parties such as cloud infrastructure or Model Providers. This does not apply where we acted intentionally or negligently.

17. Indemnity

1. You will defend us (including our staff) against third-party claims arising from any of the following, and pay the resulting losses, including reasonable legal fees:
   • Your App or Your Content;
   • a breach of these Terms, the usage policy or the law by you or End Users;
   • your failure to give End Users the required notices or to obtain their consent;
   • poor management of your account or Keys.
2. This does not cover claims to the extent they arise from our intent or negligence.
3. We will notify you of a claim without delay and cooperate reasonably. You may not settle a claim in a way that admits liability on our behalf without our consent.

18. Changes to these Terms

1. We may change these Terms within the limits of the law.
2. We announce each change in the Service at least 7 days before it takes effect, with its content, reason and effective date. For changes adverse to you, we announce 30 days ahead and also email you separately.
3. The announcement will say that you are deemed to accept unless you object by the effective date. If you do not object and keep using the Service, you are deemed to accept. If you do not agree, you may terminate before the effective date.
4. Before general availability, these beta Terms will be replaced by new terms reviewed by legal counsel, following the same procedure.

19. Notices

We send notices to your registered email or as Studio notifications. Notices to all users may instead be posted in the Service for at least 7 days. Matters with a significant effect on you are notified individually.

20. Governing law and venue

1. These Terms are governed by the laws of the Republic of Korea.
2. The parties will first try in good faith to resolve any dispute. The Suwon District Court has jurisdiction at first instance. If you are a consumer, the jurisdiction set by the Civil Procedure Act and other applicable law applies.

21. Miscellaneous

1. If any part of these Terms is invalid, the rest remains valid.
2. Our not exercising a right does not waive it.
3. You may not assign your position under these Terms without our consent. We may transfer ours in a merger or business transfer, and will notify you if we do.
4. If the Korean version and a translation differ, the Korean version prevails.

22. Company information and contact

• HIRE AI Inc. (주식회사 하이어에이아이)
• CEO: Simyung Jang
• Business Reg. No. 141-86-03506
• Mail-order business reg. 2025-용인수지-1492
• 2F 207 (No. 23), 10 Gwanggyojungang-ro 296beon-gil, Suji-gu, Yongin-si, Gyeonggi-do, Republic of Korea
• Contact: privacy@h1r.ai (during the beta this address handles support, questions about these Terms, privacy requests, and security and abuse reports)

Effective date

These Terms take effect on October 13, 2026.