Responsible AI Statement

We build AI for one of the highest-stakes settings there is. That demands more than good intentions. This statement describes the principles we hold ourselves to.

1. Human oversight, always

Trigone provides decision support, never autonomous action. The surgeon is the decision-maker. The platform surfaces information and insight, and a qualified clinician interprets and acts. We do not build features that remove the human from the loop.

2. Calibrated confidence

Confidence is treated as a first-class output. We aim to communicate certainty and doubt honestly, so an uncertain result is never mistaken for a certain one.

3. Transparency

Our insight is built to be reviewed rather than taken on faith. The platform is designed so that its outputs can be examined and understood, not delivered as unexplained verdicts.

4. Validation and generalization

We evaluate against strong external benchmarks and care about how the platform behaves on data unlike what it has seen before. A result that only holds in one setting is a hypothesis, not a capability.

5. Fairness and bias

Surgical populations, techniques, and equipment vary. We seek diverse, representative data, examine performance across groups where feasible, and treat unexplained gaps as problems to investigate.

6. Safety over automation

When a trade-off arises between doing more automatically and keeping the surgeon in control, we choose control. We would rather deliver a useful, well-bounded signal than an impressive but unsafe one.

7. Honest limitations

We document what the platform cannot do and where it can fall short. Reporting limitations openly is, to us, a feature, and the opposite of the overclaiming that erodes trust in medical AI.

8. Accountability

Because the platform is built to be auditable, its behavior can be reviewed and monitored over time. Responsibility does not end at deployment. It is a continuous obligation.

9. Privacy and consent

Responsible AI starts with responsible data. Patient videos are not permanently stored as part of routine operation, data ownership stays with our partners, and our AI improves only through carefully governed development, never by quietly retaining or repurposing customer data. See our Data Governance page.

These commitments evolve as the field and our capabilities mature. Feedback is welcome at contact@trigone.net.