Terms of Service
These terms and privacy policy are current working drafts. Your acceptance via /brief still binds under these terms as-is, but final wording may be refined during ongoing review. Questions? [email protected]
Last updated: 2026-04-18
Services and acceptance
These terms apply to every engagement starting with a submitted brief on griban.dev. Engagements under $5,000 (AI Readiness Audit, AI Automation Sprint) are accepted by checking the consent box on the brief form and receiving an email confirmation — no countersigned document required. Engagements of $12,000 and above (RAG System / Agent MVP, Full-stack AI Platform) and all monthly retainers require a separate signed service agreement before work begins.
Intellectual property transfer
Code, documentation, eval harnesses, and deliverables prepared for you transfer to you in full upon receipt of the final payment for the engagement. Before final payment you hold a non-exclusive license to use the work in production. Pre-existing libraries, boilerplates, and third-party open-source components retain their original licenses. You do not own prompts or methodology patterns that are not specific to your domain.
LLM tokens and infrastructure
All build tiers cover fixed developer labor only. LLM API usage (Anthropic, OpenAI, open-source providers), vector databases, cloud hosting, and monitoring tools are third-party costs paid directly by you. The default engagement model is bring-your-own API key: you provision your own Anthropic or OpenAI account, set usage limits, and share a scoped key with me. Managed pass-through billing is available on request for a flat $100/month administration fee.
Acceptance and quality thresholds
For every build tier, a written eval plan is produced during scoping and approved by you before implementation. The plan defines a labelled gold set (typically 100–500 examples) and three pass thresholds: faithfulness, recall, and abstention rate. A release is considered done when the code passes all three thresholds on the gold set. If a release misses threshold, I revise before invoicing. If a production regression surfaces within 30 days of handoff on a build engagement, I fix it at no additional charge.
Model deprecation and migration
When a model you depend on is deprecated by its provider, I migrate your integration to the recommended successor model within 30 days of the deprecation notice and no later than 14 days before retirement. Acceptance evals are re-run against the successor model to confirm parity; I report any observed output differences. For clients on a Production Ops retainer this is included; for clients without an active retainer, migration is quoted separately.
Limitation of liability
To the maximum extent permitted by law, my total liability arising out of or relating to any engagement is limited to the fees you have paid me in the twelve months preceding the claim. I am not liable for indirect, incidental, consequential, special, or punitive damages — including lost profits, lost data, or business interruption — even if advised of the possibility of such damages. Nothing in these terms limits liability for fraud, gross negligence, or anything that cannot be lawfully limited.
Governing law and jurisdiction
These terms are governed by the laws of Ukraine without regard to its conflict-of-laws rules. Disputes are resolved in the courts of Zhytomyr, Ukraine. At your option as a client outside Ukraine, you may alternatively bring proceedings in the courts of your principal place of business, and I agree to submit to that jurisdiction on a case-by-case basis. Nothing in this clause removes any mandatory consumer-protection right you may have under the laws of your country of residence.
How acceptance works
Accessing griban.dev and submitting a brief constitutes acceptance of these terms and the Privacy Policy. The consent checkbox on the brief form records your acceptance with a timestamp and a version string. For signed engagements you also receive a service agreement referencing the version of these terms in force at the time of signing. If these terms change in a way that materially affects your rights, I notify active clients by email with at least 30 days’ notice before the change takes effect.
Cancellation and refunds
Builds may be cancelled by either side at a milestone boundary; you pay for work delivered up to that milestone and receive everything built with a partial handoff README. Mid-milestone cancellations are billed pro-rata. The AI Readiness Audit is non-refundable once the discovery report has been delivered. Monthly retainers are month-to-month with 30 days’ written notice; unused hours in the notice month are honored, unused hours outside the notice month do not carry over.
Confidentiality
I treat all non-public information you share (business plans, code, data, user lists, strategy) as confidential and use it only to perform the engagement. A stand-alone Non-Disclosure Agreement is signed on request before any discovery work and is otherwise bundled as a clause in the service agreement for engagements above $12,000. Confidentiality obligations survive termination for five years, or indefinitely for trade secrets.
Unsolicited briefs
Submitting a brief does not create a client relationship or any obligation on my part to accept the work. I may decline any brief without stated reason. No advice given before a signed engagement constitutes legal, financial, or technical advice on which you may rely, and no confidentiality attaches to unsolicited information that is already publicly known or independently developed by me.
Changes to these terms
I may update these terms from time to time. For non-material changes (clarifications, typo fixes, new contact details) the revised terms take effect as soon as they are posted. For material changes (fee structures, liability, cancellation policy, governing law) active clients are notified by email at least 30 days before the change takes effect; the version in force at the time of signing governs the specific engagement until its conclusion.