Demo•Demo mode — synthetic data only. Do not enter real patient information.
BackAlpha — clinician oversight required

Clinical Safety

Autonomous consultation. Senior-clinician approval. COS GP independently conducts an end-to-end GP-style consultation: focused history, relevant red flags and urgency, provisional clinical reasoning and working differentials, management options, agreed next steps and personalised safety-netting. It does not action a diagnosis, prescription, investigation, referral or plan. Every conclusion and proposed action is a provisional draft requiring review, amendment or approval and sign-off by an appropriately qualified senior clinician before clinical use.

Not for emergencies.

Patients with a suspected emergency should call 999 in the UK, or 111 for urgent advice. COS GP does not provide emergency triage and must not be used to delay clinical assessment.

What COS GP does autonomously

  • Conducts the consultation end to end: focused history, red-flag exploration and urgency recognition.
  • Forms a provisional working assessment and differentials, with uncertainty stated plainly.
  • Discusses defensible management options and agrees a provisional next step with the patient.
  • Delivers personalised safety-netting in the consultation itself.
  • Surfaces clinical uncertainty and information gaps rather than hiding them.

What always requires a senior clinician

  • No diagnosis is finalised without senior-clinician approval.
  • No prescription, investigation or referral is issued by the system.
  • No management plan is actioned or entered into the record before sign-off.
  • No emergency triage and no unsupervised clinical deployment.
  • Does not replace clinical judgement or face-to-face assessment.
  • Not for emergencies and not for unsupervised clinical deployment. Evaluated on synthetic cases only.

Red flag and escalation handling

When the AI detects a possible red flag it flags it clinically in the cockpit with rationale ("why this matters"), keeps the consultation state at higher risk, and recommends clinician review before any management plan is actioned.

Audit and oversight

All AI outputs and clinician sign-off decisions are captured in an audit log scoped to the workspace. Architecture anticipates future organisation-level oversight and incident review.

Questions about clinical safety during alpha should be directed to your workspace admin.