Terms of Service

1. Service Description

SemanticGuard (the “Service”), operated by KB-AI LLC (“SemanticGuard”, “we”, “us”), is an AI gateway that provides intelligent caching, cost analytics, and request routing for Large Language Model (LLM) API calls. By creating an account or using the Service, you agree to these Terms of Service (“Terms”), the Privacy Policy, and the Acceptable Use Policy, each of which is incorporated by reference.

2. Auxiliary LLM Usage

To enable intelligent caching, SemanticGuard makes auxiliary LLM calls for template verification and entity substitution. You should be aware of the following:

• These calls are made using your API key and are billed to your account by the upstream LLM provider.
• We use the cheapest available model from the same vendor for these calls (e.g., Claude Haiku for Anthropic, GPT-4o-mini for OpenAI, Gemini Flash for Google).
• Auxiliary call volume is rate-limited per tenant per hour to prevent runaway spend. The incremental cost is typically a small fraction of the savings achieved through caching.

3. API Key Handling

• Your upstream LLM API keys are passed through to the provider at request time. They are never stored in plaintext.
• A one-way cryptographic hash of your API key is stored for identification and rate-limiting purposes.
• You are responsible for keeping your API keys secure and rotating them if compromised.

4. Caching Behavior

Responses from upstream LLM providers are cached to reduce costs and latency. Cache behavior is configurable:

• You can enable or disable caching, choose between identical-match and intelligent-caching modes, and adjust matching thresholds from the Settings page.
• Cached responses are served when a sufficiently similar request is detected, which may result in responses that are not freshly generated by the upstream model.
• You can clear the cache at any time via the dashboard or API.

5. Tier Limits and Billing

Requests exceeding your plan's limit may be throttled or rejected. You can upgrade your plan at any time. Paid plans renew automatically at the end of each billing period until cancelled. Cancellations take effect at the end of the current billing period; we do not pro-rate refunds for partial periods. Enterprise contracts have custom terms that supersede this section where they conflict.

6. Customer Data and Ownership

You retain all rights, title, and interest in the prompts, responses, and other content you submit to or receive from the Service (“Customer Data”). SemanticGuard acquires no license to use Customer Data for any purpose beyond providing the Service to you, including but not limited to training machine learning models, building benchmarks, or marketing.

You grant SemanticGuard the limited license to host, transmit, cache, and process Customer Data solely as necessary to provide the Service. This license terminates when the underlying data is deleted in accordance with your retention configuration or on account termination.

7. Service Availability

We target 99.5% monthly uptime for the hosted Service. We do not offer service credits on Free or Pro plans for periods of unavailability. Enterprise contracts may include negotiated service levels that supersede this section. The Service is designed to fail open: when SemanticGuard infrastructure is unavailable, the SDK is built to route requests directly to your upstream provider so your application continues to function.

8. Acceptable Use

Your use of the Service must comply with the Acceptable Use Policy. Violations may result in immediate suspension or termination of your account without refund. You are also responsible for the use of the Service by your end users and by any application using your SemanticGuard API key.

9. Disclaimers and Limitation of Liability

THE SERVICE IS PROVIDED “AS IS” AND “AS AVAILABLE” WITHOUT WARRANTIES OF ANY KIND, WHETHER EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, NON-INFRINGEMENT, AND ANY WARRANTY ARISING OUT OF COURSE OF DEALING OR USAGE OF TRADE.

You acknowledge that cached responses may differ from freshly generated responses and that SemanticGuard does not guarantee the accuracy, completeness, freshness, or suitability of any LLM output, whether cached or not. You acknowledge that public benchmark results published at /benchmark reflect performance on a specific test suite at a specific point in time and are not a guarantee of performance on your workload.

TO THE MAXIMUM EXTENT PERMITTED BY LAW, KB-AI LLC SHALL NOT BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, OR PUNITIVE DAMAGES, INCLUDING BUT NOT LIMITED TO LOSS OF PROFITS, DATA, OR BUSINESS OPPORTUNITIES, ARISING OUT OF OR RELATED TO YOUR USE OF THE SERVICE. IN NO EVENT SHALL KB-AI LLC'S TOTAL CUMULATIVE LIABILITY EXCEED THE GREATER OF (a) THE AMOUNT YOU PAID FOR THE SERVICE IN THE TWELVE (12) MONTHS PRECEDING THE EVENT GIVING RISE TO THE CLAIM, OR (b) ONE HUNDRED U.S. DOLLARS (US$100). These limitations do not apply to: (i) your obligations to pay fees owed; (ii) your indemnification obligations under Section 10; (iii) violations of the Acceptable Use Policy that result in direct harm to SemanticGuard or third parties; (iv) either party's gross negligence or willful misconduct; or (v) liabilities that cannot be limited under applicable law.

10. Indemnification

By SemanticGuard: we will defend you against any third-party claim alleging that the Service, as provided by us and used by you in accordance with these Terms, infringes a US patent, copyright, or trademark of that third party, and will pay damages and costs finally awarded against you in such a claim. This indemnity does not apply to: (a) combinations of the Service with materials or services not provided by us; (b) modifications to the Service made by anyone other than us; (c) use of the Service outside the scope permitted by these Terms; or (d) Customer Data.

By you:you will defend SemanticGuard and its affiliates, officers, and employees against any third-party claim arising out of (a) Customer Data, (b) your or your end users' violation of the Acceptable Use Policy, (c) your alleged infringement of any third party's rights through use of the Service, or (d) your violation of any law or regulation in connection with the Service, and will pay damages and costs finally awarded against SemanticGuard in such a claim.

The indemnified party must promptly notify the indemnifying party of the claim, give the indemnifying party sole control of the defense and settlement, and provide reasonable cooperation. The indemnification obligations in this Section 10 are not subject to the liability cap in Section 9.

11. Termination

Either party may terminate the agreement at any time, for any reason. We may suspend or terminate your account immediately and without notice for material violations of these Terms, the Acceptable Use Policy, or for non-payment.

On termination: (a) your access to the Service ends; (b) within 30 days of termination, all Customer Data, including cached responses, prompt and response content, and request traces, will be permanently deleted from our hosted infrastructure; (c) account-level records (email, hashed credentials, billing history) are retained for any period required by tax and accounting law and otherwise deleted within 90 days. You may export your data at any time before termination via the dashboard.

12. Export Controls and Sanctions

You may not use the Service if you are located in a country subject to a comprehensive US embargo (currently Cuba, Iran, North Korea, Syria, and the Crimea, Donetsk, and Luhansk regions of Ukraine), or if you are on the US Treasury OFAC Specially Designated Nationals list or any equivalent sanctions list maintained by the US, EU, or UK. You represent that none of the above applies to you or your organization.

13. Governing Law and Dispute Resolution

These Terms are governed by the laws of the State of Wyoming, without regard to its conflict-of-laws rules.

Informal resolution. Before filing a claim, you agree to try to resolve the dispute informally by emailing legal@semanticguard.dev with a description of the claim and the relief sought. The parties will use good-faith efforts to resolve the dispute within 30 days of that notice before either party initiates formal proceedings.

Arbitration.Any dispute that is not resolved informally within 30 days will be resolved by binding individual arbitration administered by the American Arbitration Association (AAA) under its Commercial Arbitration Rules. Arbitration will be held in Sheridan, Wyoming, or remotely at the arbitrator's discretion. Judgment on the award may be entered in any court of competent jurisdiction.

Class-action waiver.YOU AND SEMANTICGUARD AGREE THAT EACH MAY BRING CLAIMS AGAINST THE OTHER ONLY IN AN INDIVIDUAL CAPACITY AND NOT AS A PLAINTIFF OR CLASS MEMBER IN ANY PURPORTED CLASS OR REPRESENTATIVE PROCEEDING. THE ARBITRATOR MAY NOT CONSOLIDATE MORE THAN ONE PERSON'S CLAIMS.

Exceptions. Either party may bring an action in court for injunctive relief or for claims of intellectual property infringement. Small-claims court actions are also permitted in lieu of arbitration if they qualify.

14. Changes to These Terms

We may update these Terms from time to time. The “Last updated” date at the top reflects the most recent change. For material changes that reduce your rights or increase your obligations, we will notify active customers by email at least 30 days before the change takes effect. Continued use of the Service after the effective date constitutes acceptance of the updated Terms.

15. Compliance Status

SemanticGuard has built the technical controls SOC 2 Type II requires across the Trust Service Criteria but has not yet completed a formal audit. We make no representation or warranty of SOC 2 certification. See the Security page for our current compliance posture.

16. Contact and Notices

For questions about these Terms, contact us at legal@semanticguard.dev. Formal legal notices may be sent to that address and to KB-AI LLC, attn: Legal, 30 N Gould St Ste R, Sheridan, WY 82801, United States.