Terms of Use

These Terms govern access to and use of Tenth Man ("Company," "we," "us," "our") and the Tenth Man application, APIs, and related services (the "Service"). By accessing or using the Service, you agree to these Terms.

1. Nature of the Service

The Service is an adversarial decision intelligence platform that generates structured analytical outputs using artificial intelligence systems. The Service is specifically designed to surface dissent, challenge assumptions, and identify blind spots in high-stakes decisions through institutionalized disagreement.

The Service uses a three-agent architecture:

• Strategist: Generates an initial recommendation with explicit assumptions, constraints, and risks.
• Skeptic (Tenth Man): Performs mandatory adversarial critique, including incentive misalignment, second-order effects, and blind spots. The Skeptic is structurally required to disagree and will attack the strongest argument in favor of the recommended action.
• Synthesizer: Arbitrates between agents, picks a side, explicitly rejects weak arguments, and issues a final recommendation with calibrated confidence.

Because the Service is designed to produce deliberate disagreement between agents, outputs from different agents will often contradict each other. This is by design, not a defect. The final Synthesizer output reflects structured arbitration, not consensus.

The Service does not act autonomously, execute decisions, or monitor real-world outcomes. It is a decision-support tool only.

2. Adversarial Output and Institutionalized Dissent

You acknowledge and agree that:

• The Service is expressly designed to challenge, contradict, and critique the prevailing view on any question submitted.
• Agent outputs will frequently conflict with each other as a function of the system's architecture.
• The Skeptic agent is structurally required to attack the strongest argument in favor of the recommended action, even when that recommendation may ultimately be correct.
• Dissent produced by the Skeptic does not indicate system error and should not be dismissed solely because it conflicts with your prior assumptions or the Strategist's recommendation.
• The Synthesizer's final recommendation reflects the system's arbitrated judgment based on submitted information only, not independent verification of facts or claims.

Nothing in the Service's output constitutes an endorsement, rejection, or validation of any particular course of action.

3. No Professional Advice; No Reliance

The Service does not provide legal, medical, investment, financial, accounting, tax, regulatory compliance, safety, engineering, or fiduciary advice. Outputs do not constitute professional advice and must not be treated as such. You will independently evaluate all outputs and will not rely on the Service as a substitute for licensed professionals.

4. Immutable Audit Trail

The Service produces decision artifacts that are append-only and write-once by architectural design:

• Decision outputs, comparison artifacts, classification metadata, memory declarations, and evidence snapshots are stored immutably once written.
• Artifacts are never retroactively modified by the system.
• Each artifact contains a canonical decision ID, creation timestamp, and full agent outputs for audit purposes.
• Where follow-up decisions are run, the memory declaration submitted by the user is stored verbatim as an immutable provenance record.

This immutability is a system integrity guarantee, not a promise of accuracy. Immutable storage reflects what the system produced and what you declared; it does not validate the correctness of either.

5. Explicit Continuity; No Inferred Memory

The Service does not infer context, history, or resolution between decision runs. When running a follow-up decision, you are required to explicitly declare: assumptions carried forward from the prior decision; new information added since the prior decision; objections claimed as resolved; and objections still open.

This declaration is accepted verbatim. The system does not review, validate, or augment your declaration. You are solely responsible for the accuracy and completeness of all memory declarations submitted. If you claim that an objection has been resolved when it has not, or omit material new information, the resulting output will reflect those omissions without correction.

6. Confidence Calibration; Not a Statistical Guarantee

Confidence scores produced by the Service are internally derived structural constraints, not statistical guarantees of outcome probability:

• Confidence is hard-capped when the number of unresolved uncertainties exceeds defined thresholds.
• A higher confidence score indicates fewer unresolved uncertainties at decision time, not a higher probability of success.
• Confidence scores will be compressed in scenarios involving catastrophic, irreversible, or externally harmful risk where mitigation is incomplete.
• Confidence scores do not account for information you failed to provide, facts outside the system's evidence base, or events occurring after the decision was run.

7. Responsibility Allocation

You retain sole responsibility for: all decisions made using the Service; all actions taken based on outputs; compliance with applicable laws and regulations; human oversight and review of all outputs before acting; and the accuracy and completeness of all inputs and memory declarations submitted. The Service does not assume decision authority. The fact that the system produced a recommendation does not transfer any portion of decision responsibility to the Company.

8. Regulated and High-Risk Use

The Service is not designed, intended, or certified for use in medical diagnosis or treatment, clinical decision-making, emergency response systems, life-support or safety-critical systems, aviation or defense control systems, nuclear facility operations, automated law enforcement, financial trading execution systems, or any use requiring government certification or regulatory approval. Use in such environments is at your own risk and subject to your independent compliance obligations. Nothing in these Terms authorizes use in violation of applicable law.

9. AI System Limitations

You acknowledge that:

• AI systems can generate inaccurate, incomplete, biased, or misleading outputs.
• The Skeptic agent is designed to attack the prevailing recommendation aggressively; this attack may be factually incorrect, overstated, or based on assumptions not supported by submitted evidence.
• The Synthesizer may resolve agent disagreement in ways that do not reflect the full weight of available evidence.
• Risk classifications, severity labels, and harm domain categorizations are structured approximations, not authoritative assessments.
• External web evidence, where enabled, may be incomplete, outdated, or inaccurate.
• The system does not guarantee accuracy, completeness, timeliness, or correctness of any output.

10. External Evidence Disclaimer

When web evidence retrieval is enabled:

• Third-party content is retrieved and stored as immutable snapshots at the time of the decision run.
• Evidence snapshots are frozen at retrieval time. The system does not update snapshots to reflect subsequent changes in third-party content.
• All agents in a given decision run operate on the same frozen evidence snapshot.
• We do not control, verify, or warrant the accuracy of third-party content.
• Search retrieval is not exhaustive. Relevant sources may be omitted.
• You remain responsible for independent verification of any third-party information that is material to your decision.

11. No Warranties

TO THE MAXIMUM EXTENT PERMITTED BY LAW, THE SERVICE IS PROVIDED "AS IS" AND "AS AVAILABLE." WE DISCLAIM ALL WARRANTIES, INCLUDING WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, NON-INFRINGEMENT, ACCURACY, RELIABILITY, SECURITY, AND AVAILABILITY. We do not warrant uninterrupted or error-free operation.

12. Limitation of Liability

TO THE MAXIMUM EXTENT PERMITTED BY LAW, IN NO EVENT SHALL COMPANY BE LIABLE FOR INDIRECT, INCIDENTAL, CONSEQUENTIAL, SPECIAL, OR PUNITIVE DAMAGES, LOST PROFITS, LOST REVENUE, BUSINESS INTERRUPTION, LOSS OF DATA, REGULATORY PENALTIES, OR PERSONAL INJURY OR DEATH, ARISING OUT OF OR RELATED TO USE OF THE SERVICE.
Aggregate liability shall not exceed the total fees paid by Customer in the 12 months preceding the claim, or $100 if no fees were paid. Some jurisdictions may limit enforceability of certain exclusions.

13. Indemnification

You agree to indemnify and hold harmless Company from any claim arising out of: your use of the Service; your decisions and actions based on Service outputs; your regulatory obligations; your submitted content and memory declarations; or your violation of these Terms.

14. Data, Logging, and Artifacts

The Service stores decision artifacts, rerun provenance records including verbatim memory declarations, evidence snapshots frozen at retrieval time, comparison and classification artifacts, and telemetry logs. Artifacts are immutable and append-only for system integrity purposes. You represent that you have rights to all content submitted. Data handling practices are governed by the Privacy Policy.

15. Export Controls and Sanctions

You may not use the Service in violation of U.S. export control laws, sanctions regulations, or trade restrictions. You represent that you are not located in a sanctioned jurisdiction and are not on any denied or restricted party list.

16. No Agency; No Fiduciary Duty

Use of the Service does not create agency, partnership, joint venture, fiduciary duty, or any advisory relationship. The Service is software, not a human advisor. The adversarial structure of the Service's outputs does not create any duty of care beyond what is expressly stated herein.

17. Acceptable Use

You may not use the Service to facilitate illegal activity, plan harm, bypass regulation, generate disallowed content, reverse engineer proprietary systems, or abuse usage limits. We may suspend or terminate access for violations.

18. Dispute Resolution and Arbitration

Disputes shall be resolved via binding arbitration. You waive rights to jury trial and class action participation. Arbitration venue and governing law: State of Delaware, USA.

19. Changes to Service or Terms

We may modify features, pricing, or these Terms at any time. Continued use constitutes acceptance of the modified Terms.