Evaluation Terms

Effective date: October 26, 2026
Provider: nudgIQ Inc., doing business as Tier3 Labs (“Tier3 Labs”, “we”, “us”, “our”).

These Evaluation Terms apply to any sample tasks, environments, graders, trajectories, difficulty manifests, build and grading documentation, benchmark results and other materials we provide to you or your organization (“you”) for evaluation before a signed agreement is in place (together, the “Evaluation Materials”). By accessing or using Evaluation Materials you agree to these terms on behalf of yourself and the organization you represent. If we sign a separate evaluation agreement, non-disclosure agreement or order form with you, that document controls to the extent it conflicts with these terms.

1. What we provide

We will tell you in writing, usually in the message that delivers the Evaluation Materials, what is included, how to access it, and the evaluation period. Unless we say otherwise, the evaluation period is sixty (60) days from delivery. We may extend it on request and may end it early on notice if we believe these terms have been breached. We may provide support during the evaluation at our discretion, and we are not obliged to update or maintain Evaluation Materials.

2. License

During the evaluation period, we grant you a limited, non-exclusive, non-transferable, non-sublicensable, revocable license to use the Evaluation Materials internally, for the sole purpose of evaluating whether to license our environments and services. This includes running the sample tasks against your models, inspecting the graders and documentation, and producing internal reports and analyses of the results.

3. Restrictions

Except as we expressly permit in writing, you will not, and will not allow anyone else to:

• train on it: use the Evaluation Materials, any trajectories, rewards, grader outputs or reference solutions derived from them, or any data generated by running them, to train, fine-tune, distil, align or otherwise improve any model intended for production, commercial or external use;
• copy or extract: copy the Evaluation Materials except as needed to run them, or extract tasks, fault scripts, checks, rubrics, reference solutions, seeds or world data for use outside the evaluation;
• build on it: use the Evaluation Materials to design, build or improve a product, dataset or service that competes with ours, or to build environments that reproduce our tasks, worlds or grading methodology;
• disclose it: share the Evaluation Materials or any non-aggregate results with anyone outside your organization, including contractors not bound by equivalent confidentiality obligations, or publish or publicly disclose benchmark results without our prior written consent;
• circumvent: disable, probe or work around the logging proxy, call budgets, grader isolation, scope checks or other controls, except as part of a red-team exercise we have agreed to in writing;
• strip notices: remove or alter any proprietary, license or attribution notice; or
• misuse the substrate: use the open-source software included in an environment in a way that breaches its own license.

Nothing in these terms restricts your use of your own models, prompts or infrastructure, or of general knowledge and skill that your personnel retain in unaided memory.

4. Ownership

The Evaluation Materials are our proprietary work product and are licensed, not sold. We and our licensors retain all right, title and interest in them, including every environment, world, seed, fault script, task, check, rubric, reference solution, grader, harness, trajectory format, manifest and document, and in any modifications or derivatives, whoever makes them. You own the models you evaluate, the prompts you write, and your internal reports. Trajectories and grader outputs produced by running the Evaluation Materials may be used by you only as part of the evaluation under section 2.

Open-source software included in an environment remains under its own license, which we retain and pass on unmodified. Third-party product names that appear in the Evaluation Materials belong to their respective owners, who do not sponsor or endorse us.

5. Confidentiality

The Evaluation Materials, their contents, pricing, proposals, and any non-public information we share about our methods, roadmap or customers are our confidential information. Information you share with us about your models, evaluation plans, results and infrastructure is your confidential information. Each party will protect the other’s confidential information with at least the care it uses for its own information of similar sensitivity, use it only for the evaluation, and share it only with personnel and advisers who need it and are bound by confidentiality. Confidential information does not include information that is public without breach, was already known to the recipient, was received lawfully from a third party, or was developed independently. These obligations last for three years after the evaluation period ends, and indefinitely for trade secrets.

6. Provenance and data

The worlds inside our environments are fictional: companies, people, tickets, records and configuration are generated from a fixed seed, and generated names are checked against the public record so that they do not refer to real organizations. We do not use real customer data, scraped support tickets or other personal data to build them. Environments are built on unmodified, version-pinned open-source products used within their licenses. We will, on request, provide a written statement of provenance and a list of the open-source components and licenses in any Evaluation Materials.

You will not introduce real personal data, customer data or confidential third-party data into an environment during the evaluation.

7. Feedback and results

We welcome your observations about task quality, grader behavior, difficulty and packaging. You grant us a perpetual, irrevocable, royalty-free license to use feedback you give us to improve our environments and services, without attribution and without obligation. We will not publicly identify you or your organization as an evaluator, or publish results of your evaluation, without your written consent.

8. End of the evaluation

When the evaluation period ends, or earlier if we ask, you will stop using the Evaluation Materials and, within ten business days, delete or return all copies, including derived trajectories, grader outputs and extracted content, except for one archival copy of your internal reports and any copies retained automatically in backups for a limited period, which stay subject to section 5. On request you will confirm in writing that you have done so. Sections 3 through 10 survive the end of the evaluation.

9. Warranties, liability and indemnity

The Evaluation Materials are provided “as is” for evaluation only. To the fullest extent permitted by law we disclaim all warranties, express or implied, including merchantability, fitness for a particular purpose, non-infringement, and that the Evaluation Materials are free of errors, that pass rates will match those we report, or that results will be reproducible on your infrastructure. The warranties, acceptance criteria and remedies that apply to a commercial engagement are those stated in the applicable signed agreement.

To the fullest extent permitted by law, neither party will be liable to the other under these terms for indirect, incidental, special, consequential, exemplary or punitive damages, or for lost profits, revenue, data or goodwill, however caused. Each party’s total liability under these terms will not exceed one thousand US dollars (US$1,000). These limits do not apply to a breach of sections 3 or 5, misuse of the other party’s intellectual property, or liability that cannot be limited by law.

You will indemnify us against claims, losses and expenses (including reasonable legal fees) arising from your breach of section 3 or 5 or from data you introduce into an environment.

10. General

These terms are governed by the laws of the State of Delaware, United States, without regard to conflict-of-law rules, and any dispute will be brought exclusively in the state or federal courts located in Delaware; either party may seek injunctive relief anywhere to protect its intellectual property or confidential information. Each party acknowledges that a breach of section 3 or 5 may cause harm that damages cannot adequately remedy. You may not assign these terms without our consent. If any provision is unenforceable, the rest remains in effect. These terms, with our Terms of Service and Privacy Policy, are the entire agreement about the evaluation, and may be modified only in a writing signed or expressly agreed by both parties. We may update these terms for future evaluations by posting a revised version here; the version in effect when you received your Evaluation Materials governs that evaluation.

Contact

nudgIQ Inc., doing business as Tier3 Labs