Everything you might be wondering.

Elderbot is a therapist-driven OT-AI platform. A small set of privacy-preserving home sensors and a friendly chatbot notice the patterns of daily life. Our AI turns those signals into recognised activities and a daily timeline, then gives occupational therapists a clear read — summaries, deviation alerts and evidence support — so they can act between visits.

No — never. Privacy is a design principle, not a setting. Elderbot recognises activity from acoustic and ambient signals — running water, footsteps, a chair moving, a door opening — not video and not conversation transcripts.

The sound sensor listens for acoustic events, not speech. On-device models classify sounds like a toilet flush, utensils, or a tap running — and discard the rest. We store the event ("water running, 07:12"), not an audio recording of your home.

Core daily activities (ADLs) including bathing, transfers, toileting, feeding and meal preparation, and medication routines — plus fall detection. Each maps to the Person–Environment–Occupation (PEO) framework therapists already use.

No. It's built to keep the therapist in the loop. The AI surfaces patterns, alerts and evidence; the therapist interprets, decides and adjusts the intervention. Elderbot gives clinicians more of what they lack — continuous visibility and time — never less authority.

Occupational therapists working in NGOs and private clinics, and the service providers and funders who coordinate elderly care. Therapists get decision support; providers get population-level micro-need intelligence.

A sound sensor (the primary sensor), plus supporting pressure, contact and floor-vibration sensors, an optional wearable, and a care chatbot. No cameras, and nothing that requires the older adult to operate complex technology.

Processing stays on-device wherever possible, we keep derived events rather than raw audio, and access is limited to the person's care team. Elderbot is designed around data minimisation — capture the least needed to help, and nothing more.

Elderbot is benchmarked against gold-standard measures — the Barthel Index, Lawton IADL, MoCA and COPM — and validated through focused pilot studies on bathing, transfer and fall detection, evaluated for feasibility, acceptability and preliminary clinical value.

Beyond activity analytics, an agentic evidence assistant searches and summarises relevant research, then presents it against the therapist's clinical focus. The therapist selects the evidence that fits — decisions stay evidence-informed and human-led.

Tell us about your caseload and goals. We scope a small, therapist-led pilot, install the sensing layer, and review what it surfaces together. Request a demo →