Sevah Robotics — robotics for healthcare — 2026

Imagine healthcare
where the care staff are at the bedside.

We call it care augmentation. Pair the people in scrubs with a robot that can fetch, watch, draft, and remember — and the work that only a human can do gets the time it deserves.

Sevah builds the robots, the orchestration, and the quiet bedside intelligence behind it: a Companion in the room, a fleet in the hallway, a steady hand in the chart.

Scroll

The chart

Today

Care staff end a twelve-hour shift with two more hours of charting waiting at the keyboard. Most of what gets written is written from memory.

With Sevah

The note is already drafted by the time the nurse sits down — timestamped, sourced, and waiting for a signature. The two hours go home.

The hallway

Today

A bedside aide walks 14,000 steps a shift, half of them pushing a cart that weighs more than the patient waiting for it. The bedside is empty.

With Sevah

The cart drives itself. The aide is in the room with the patient who needed them, doing the work they were hired to do.

The handoff

Today

Shift change is twenty minutes of verbal handoff and a printed sheet. Half of what mattered last night is already lost by morning rounds.

With Sevah

The handoff is a structured summary built from the room, not from a tired memory. What mattered last night is the first thing the day shift sees.

The intake

Today

An intake nurse asks the same eight questions every patient has answered four times already. The conversation is a form being filled, not a person being heard.

With Sevah

The form is already populated. The first thing the patient hears is a question only a human would think to ask.

The audit

Today

The week before a regulator visits is a week of midnight panic. Records are reconstructed from memory. Citations are written from gaps.

With Sevah

The auditor opens the record and finds a trail — who observed what, when, signed by whom. The walk-through ends before lunch.

The family call

Today

The daughter calls every day at 4pm. The nurse who knows the patient is in another room. The voicemail comes back tomorrow.

With Sevah

The call is answered. The summary is read back from the patient's day. When a nurse joins, only the part that needed a nurse is left.

The supply

Today

A nurse leaves a room mid-task to find a missing kit. By the time they return, the moment has shifted and so has the patient.

With Sevah

The kit arrives before it is asked for, because the room said it would be needed. The nurse never had to leave.

The work itself

Today

Healthcare is among the leading professions by burnout in America. The reasons most care staff give for leaving have nothing to do with the patient.

With Sevah

The reasons to stay come back. The conversation. The touch. The room. The judgment that the license was given for in the first place.

Care augmentation is what Sevah builds: a robot that takes the load a person should never have carried, so the person can do the part only a person can. Wherever care happens.

Introducing

CareOS.

An AI operating system that captures patient needs between visits, dispatches staff and robots, and turns every care interaction into structured documentation.

100%

Nurse-signed

Units. One console.

90d

One wing. No capex.

The Sevah Principle

सेवा

Sanskrit · selfless service · the act of giving care without expectation of return

“Give first. Help without expectations.”

Give first.

Within the bounds of consent and the facility's HIPAA framework, a Sevah unit works ahead of the request. The system listens, drafts, fetches, and watches — so the help is already on its way by the time anyone needs to ask.

The resident comes first.

Every alert, every drafted note, every supply run exists for one reason: to put more of the right care in front of the right person at the right time. Care staff get their attention back. Residents get a caregiver who isn't buried.

We streamline the chart.

Charting is the second shift no one signed up for. Sevah listens through the day, drafts the observation in the moment, and hands the nurse a finished note ready for review and signature. The nurse gets two hours of their life back.

No false positives.

The wrong page is worse than no page. Every alert clears a confidence threshold and a severity gate before it reaches a human. We page the nurse when something has actually changed — and we earn the trust that the next page is real.

Not marketing copy. An engineering constraint. Every model, every product decision, every revenue mechanic is checked against these four rules before it leaves the building.

01 / The Problem

Three quarters of charting is observation. We accelerate that part.

A nurse's job is to clinically assess. The chart isn't a transcript — it's the record of assessment, built from a thousand small observations: the half-finished glass on the nightstand, the IV bag that needs a refill, a slower gait, a slight hoarseness in speech. Sevah captures those signals continuously and feeds them up as structured observations the nurse can confirm and turn into clinical judgment.

charting time · the 75/25 split

75% OBSERVATION

25% CLINICAL

◀ Sevah drafts this · auto-attributed, RN-signednurse owns this ▶

What Sevah unlocks · by role

CNA1 : 10

Stays on the floor instead of running supply carts — more hands for residents, fewer errands.

LN1 : 20

Walks into every room already informed — observations drafted, hydration flagged, early deterioration surfaced.

RN1 : 30

Leaves the shift with the chart largely drafted, every observation attributed to source — two hours returned to care.

02 / The Lineup

Two robots. One operating system.

Each unit is built for a different part of the day, but they share one runtime and one console. Start with one. Add the next when the first has earned the right to ask.

Phase 01 · Available now

Companion

The calm listener on every nightstand.

For the resident who would otherwise be alone.

A matte-ceramic unit that lives at the bedside, listening more than it speaks. It captures requests, watches for changes in gait and mood, prompts a sip of water when one is overdue — and when something feels off, it pages the charge nurse before a fall, not after.

One per bedFalls · call-light responsePer resident-month

Phase 02 · 1–2 years

Orchestrator

Silent supply, sized for hallways.

For the nurse aide who shouldn't be pushing a cart all shift.

Linens, briefs, PPE, meal trays — the work that pulls a CNA off the floor for hours at a time. Orchestrator glides room to room on a LiDAR route it shares with Companion, so an entire shift of hauling can come off your team.

1–3 per wingCNA floor-time · stockoutsPer wing-month

Phase 01 · Available now

CareOS

The operating system beneath every robot and every workflow.

For the facility that needs one console, not twelve integrations.

CareOS listens to what Companion and Orchestrator observe, cross-references care schedules and staff assignments, and dispatches the right action — a page, a supply run, a drafted note — to the right person. One runtime. One audit trail. Every robot, every shift.

EHR-agnostic · API-firstSingle pane · all robotsPer facility-month

03 / How We Land

We earn the chart before we touch it.

We begin with the workflows that hurt most — the hauling, the prompting, the request-taking — and we only move closer to the chart once your team has decided we should.

Workflow Compression

Phase 01 · Now

Companion takes up residence at the bedside — listening, prompting, watching for falls and deterioration signals, paging the right person at the right time.

Charting Accelerant

Phase 02 · Q3 2026

Observations gathered through the day arrive at the nurse station as drafts, with sources attached, ready for a clinician to read, edit, and sign.

Inbound / Outbound Calls

Phase 03 · 2027

Family calls, insurer follow-ups, scheduling — routed, transcribed, summarized — and the switchboard quietly stops costing a full-time equivalent.

04 / A Shift

What your floor looks like with Sevah running.

A 7a–7p shift. 42 residents. One charge nurse, four CNAs. Sevah runs in the background — three products, one console. Only the exceptions surface.

Companion · 42 rooms