AI care
for oncology.
Voice-first AI care for oncology. Silver listens, reads the body through wearables, stays in sync with the treatment calendar, and answers quiet questions, always under the oversight of a real clinician. A companion for the patient. A signal for the care team.
Between appointments,
symptoms go unseen.
Oncology patients spend two to three weeks between visits. During that interval, toxicities evolve — nausea, fatigue, neuropathy, pain — but there is no continuous channel back to the care team. By the next appointment, recall is incomplete, severity is underestimated, and clinically actionable signals are lost.
Basch et al., JAMA 2016 · Basch et al., JCO 2017 · Di Maio et al., NEJM 2022
Closing this gap saves lives. Multiple randomised trials have shown that when patients report symptoms electronically between visits, clinicians detect toxicities earlier, emergency admissions drop, and overall survival improves by up to five months.
The tools that proved this — PRO-CTCAE, ESAS, structured ePRO questionnaires — share a common limitation: adherence declines over time. The forms are repetitive, clinical in tone, and disconnected from the patient’s daily experience. Completion rates fall as treatment fatigue rises, precisely when longitudinal data matters most. The evidence is settled. The problem is sustained engagement.
Silver lives in
that gap.
Three senses, a calendar, and a clinician. Silver accompanies the patient every day, notices what the care team can’t, and hands over the moment a specialist is needed.
Voice
Daily check-ins by voice. Ninety seconds, no forms, no scales. The companion speaks the patient's language, listens, and extracts the clinical signal silently.
Body
Unified wearable data from Apple, Google, and Whoop. Sleep, heart rate, activity, and more — surfaced as a single recovery score.
Words
Text Q&A for the quiet questions about treatment, symptoms and side effects. Grounded in the patient’s actual data, not a generic chatbot.
A clinician
A specialist always on the other side of the screen. The AI handles the everyday; the clinician steps in on red flags, on uncertainty, and whenever the patient asks for it.
A visual language
for being ill in.
Every mark, colour and line was chosen to lower the cost of being here. No clinical blue. No sharp edges. No loud anything. The identity is the therapy.
Silver draws on principles from architecture and perception research: spaces that adapt to who arrives (Alexander), interfaces the body understands before the mind parses (Norman), colour as relationship not decoration (Albers), the effort of existing as a clinical variable (Sternberg), and silence as a generous act (Zumthor).
A presence, not a button.
The hero element.
The orb breathes at a normal resting respiratory rate. It is always on screen, never hidden behind a menu. A presence, not a button.
Four ways in,
one companion.
The orb is the hero. Every other surface is a room off the main hall, entered only when the patient has the bandwidth.
The moment a
clinician steps in.
Silver does most of the everyday work alone: the gentle check-ins, the symptom tracking, the questions about treatment. But the AI knows where its edges are. A five-level escalation system runs on every encounter. Every agent output passes through safety checks before reaching the patient. When something crosses a line, Silver acts with full context via a structured clinical handoff to the care team.
Patient directed to hospital.
When the system detects a potentially life-threatening adverse event — based on predefined clinical criteria reviewed by the supervising oncologist — it instructs the patient to seek emergency care immediately, in their own language, and simultaneously alerts the care team with full clinical context.
- Immediate, unambiguous instruction to the patient
- Structured clinical handoff to the oncology service
- Companion stays with the patient through the transition
- Encounter tagged as IMMEDIATE · hospital-directed in the audit trail
Triage thresholds follow published oncology emergency guidelines. The specific criteria for each condition are defined in the clinical protocol and reviewed by the supervising oncologist.
Clinician notified within the hour.
Grade 3 symptoms that don’t demand the ER but do demand attention. Silver pushes a structured notification with the symptom, the trend, the wearable context, the cycle day, and the AI’s recommended next step. The clinician decides.
Built for regulated health data.
Silver processes sensitive health data in compliance with applicable privacy frameworks — GDPR in Europe, HIPAA in the US, LGPD in Brazil, and equivalent regulations worldwide. Voice recordings, wearable signals and clinical symptoms are stored encrypted, with explicit patient consent, and retained only for the duration of the care relationship. Patients can request access, correction or deletion at any time. Data residency is configurable per deployment.
Consent
Explicit informed consent before any data collection. Patients can withdraw at any time without affecting their care.
Storage
Data residency follows local regulation. Encryption at rest, transmitted over TLS. Retention limited to the active care period plus the legally required archive.
Access rights
Patients can view, export, correct or delete their data. Clinicians access only their own patients. Audit trail on every access.
AI Care,
under clinician oversight.
Not a chatbot. Not a surveillance system. AI care that listens, watches over, stays in sync with the calendar, and knows when to call the specialist.