Conefia
- 1 follower
- United States of America
- https://conefia.com
- company/conefia
- conefia.tech
- dev@conefia.com
Popular repositories Loading
-
SAFE-CARE
SAFE-CARE PublicSAFE-CARE™ — a safety and escalation framework for healthcare AI agents. Defines the capability layers (retrieval, citation, guardrails, session memory, escalation) and the standard for what a safe…
Python
-
SAFE-CARE-Bench
SAFE-CARE-Bench PublicReproducible benchmark for multi-turn healthcare AI agent safety. A0-A4 controlled ablation isolates what each layer - retrieval, guardrails, memory - actually contributes. 8 scored dimensions, sin…
Python 1
-
SAFE-CARE-SDK
SAFE-CARE-SDK PublicOfficial open-source SDK for the SAFE-CARE™ framework — a thin, provider-agnostic client for running SAFE-CARE safety evaluations and wiring escalation guardrails into healthcare AI agents. Connect…
-
blind-by-construction
blind-by-construction PublicCan a judged benchmark see the safety control it evaluates? Ours left a third of a new guardrail’s actions unscored, under every judge. Here are the scenarios, every scored row, a physician’s label…
Python
Repositories
- SAFE-CARE-Bench Public
Reproducible benchmark for multi-turn healthcare AI agent safety. A0-A4 controlled ablation isolates what each layer - retrieval, guardrails, memory - actually contributes. 8 scored dimensions, single & multi-turn escalation judges, hard-fail override.
- SAFE-CARE Public
SAFE-CARE™ — a safety and escalation framework for healthcare AI agents. Defines the capability layers (retrieval, citation, guardrails, session memory, escalation) and the standard for what a safe conversational health agent must do and must never do. Sole-authored, patent-filed. Instrumented by SAFE-CARE Bench.
- blind-by-construction Public
Can a judged benchmark see the safety control it evaluates? Ours left a third of a new guardrail’s actions unscored, under every judge. Here are the scenarios, every scored row, a physician’s labels and a check for each of five defects.
- SAFE-CARE-SDK Public
Official open-source SDK for the SAFE-CARE™ framework — a thin, provider-agnostic client for running SAFE-CARE safety evaluations and wiring escalation guardrails into healthcare AI agents. Connects to the hosted SAFE-CARE evaluation service; implements no proprietary scoring itself.
People
This organization has no public members. You must be a member to see who’s a part of this organization.
Top languages
Loading…
Most used topics
Loading…