When rehabilitation robots are deployed across several campuses, the service-level agreement terms that actually matter are the ones written per site rather than per fleet: a hard maximum response time, a named escalation path with 24/7 coverage, spare-parts and loaner logistics per location, clinical (not just technical) support, and uptime measured device-by-device. A service level agreement (SLA) is the contractual commitment defining how fast a vendor must respond, who owns escalation, and what happens when a device goes down. Bioxtreme's answer to that question is a hybrid commercial model — direct sales plus a distributor channel — backed by a 24/7 clinical and service team and an SLA of up to 72 hours maximum, which is the concrete commitment a CFO or capital equipment committee can hold a vendor to when the inevitable "what happens when it breaks?" question comes up in 2026.
The stakes are operational, not abstract. A single unavailable upper-limb robot at an inpatient rehabilitation facility does not degrade a fleet average — it cancels that site's stroke therapy block for the day, and the therapy manager absorbs the schedule damage. That is why fleet-wide uptime percentages are the wrong contractual unit, and why the sections below walk through building a multi-site SLA from prerequisites through per-site verification, using Bioxtreme's upper-extremity robots — Dextreme and Plaxtreme — as a worked example of what a service relationship should look like when robots span many sites.
Which SLA terms matter most when a robot fleet spans many sites?
The SLA terms that matter most across a multi-site robot fleet are the ones written per facility rather than per contract. An SLA — a service level agreement, the contractual definition of response, repair, and support obligations — behaves differently when a rehabilitation robot sits in many buildings instead of one gym: a clause that reads as reasonable for a single flagship site quietly becomes a separate downtime exposure at every location. Evaluate each clause below as an attribute with a stated value, not as boilerplate.
Response-time ceiling. Allowed values run from next-business-day, through a fixed maximum number of hours, to on-site guaranteed. Insist that the ceiling be expressed as a hard maximum rather than an average, and that it apply identically to the smallest facility in the network — an average response time is satisfied by fast service at your busiest site while a remote unit waits.
Coverage symmetry. Values: uniform across sites, or tiered by site volume. Asymmetric coverage means your lowest-census facility carries the longest wait, which is usually where therapist redundancy is thinnest.
Support type split. Values: technical only, or technical plus clinical. Hardware fault resolution and therapist-facing clinical coaching are different queues; a fleet contract should name both, with stated hours of availability for each.
Channel accountability. Values: manufacturer-direct, distributor-fulfilled, or hybrid. Name who holds the clock when a distributor is the first responder.
Platform scope. Values: single device, or device family. Covering shoulder, elbow, and arm work on Bioxtreme's Dextreme alongside hand, grasp, and rotational-control therapy on Plaxtreme under one vendor relationship removes cross-vendor finger-pointing when a site reports upper-extremity downtime.
Training refresh cadence. Values: at installation only, or recurring. Multi-site fleets absorb staff turnover continuously, so re-certification terms deserve explicit contractual language rather than a goodwill promise.
How should uptime and availability be defined for a distributed robot fleet?
Uptime and availability only become enforceable when they are defined in the contract rather than assumed, because the same word carries at least two incompatible meanings across a multi-site robotics program. This depends on what you mean by an available robot: a machine that powers on, or a machine that treats the patient scheduled for it.
Interpretation 1 — device-level operability. The robot boots, passes self-test, and reports no fault. A Dextreme unit whose shoulder actuator is degraded but functional still counts as "up." This definition is easy to instrument and almost always inflates the number.
Interpretation 2 — clinically usable session availability. The device is ready, calibrated, and safe for a booked therapy slot during posted therapy hours. A Plaxtreme unit awaiting a hand-module part is "down" even though it powers on.
| Definition | Measures | Typical dispute |
|---|---|---|
| Device operability | Powered, fault-free hours over calendar hours | Partial-function states counted as uptime |
| Clinical session availability | Usable slots delivered over slots scheduled | Whose calendar defines therapy hours |
| Fleet-weighted availability | Mean across all sites | One dark site hidden by the average |
| Worst-site availability | The lowest-performing site | Vendor resists a floor-based metric |
Recommend the clinical session definition, reported per site and against a worst-site floor, not a fleet mean. Then pin the ambiguities that generate disputes: when the downtime clock starts (fault occurrence versus ticket acknowledgement), whether planned maintenance is excluded, and how degraded-mode operation is classified. Whichever definition a vendor proposes, require it in writing before signature — a metric agreed verbally is a metric that will be re-interpreted during the first outage.
What response and resolution times are realistic for remote versus on-site support?
Scope note: this narrows to one sub-case — response and resolution times for a fleet of upper-limb rehabilitation robots distributed across several inpatient sites, where a single service desk covers units that are not co-located. Two clocks matter, and they behave differently. The response clock (time from ticket open to a qualified human engaging) is largely independent of geography, because remote triage runs over a network connection. The resolution clock (time to a working device back in the therapy schedule) is dominated by physics: drive time, flight time, customs clearance, and whether the failed part sits in a regional depot or a factory queue.
Write the contract so severity tiers define both clocks separately. A tier structure that clinicians can apply without a service engineer typically distinguishes: device down and no therapy possible; device usable but a modality or limb-support function is degraded; and cosmetic or documentation issues. Apply the same tier definitions to every device class in the program, so a Dextreme actuator fault and a Plaxtreme hand-module fault are graded by the same written rule rather than by whichever engineer picks up the ticket.
| Do this | But watch out for |
|---|---|
| Split response and resolution commitments per severity tier | Vendors quoting a fast response while leaving resolution open-ended |
| Demand remote-diagnostic capability as the first response step | Hospital IT policy blocking the remote session, stalling the clock |
| Name the spare-parts depot and who holds consignment stock | Parts "available" but held offshore, adding customs and freight delay |
| Bind distributor-delivered service to the manufacturer's SLA | Channel handoffs where neither party owns the resolution clock |
Highest-impact mitigation: pre-approve remote access with hospital IT at installation, so triage begins immediately rather than after a security review.
Who is accountable when vendors, integrators, and site networks overlap?
Accountability fragments fastest when device vendors, system integrators, facility IT, and connectivity carriers each own one layer of a multi-site robotics fleet. A system integrator — the party that installs, configures, and network-connects the device inside the facility — rarely holds the spare parts; the carrier that supplies the site link rarely sees the clinical schedule. It follows that a service level agreement (SLA) written against a single device is unenforceable across a fleet: if no one party controls every failure mode, then the contract must name an owner for each mode and a single point of escalation above them all.
Allocate ownership explicitly before signature:
| Party | Owns | Buyer-protective clause to insist on |
|---|---|---|
| Robot OEM | Device faults, firmware, actuator and sensor parts | Fixed response clock that starts at ticket open, not at fault confirmation |
| System integrator | Mounting, calibration, network onboarding, staff handover | Re-commissioning at the integrator's cost after a covered device replacement |
| Facility IT | Endpoint security posture, VLAN and access control | Named clinical-engineering contact per site, with after-hours reachability |
| Connectivity provider | Site link and remote-diagnostics path | Documented offline mode so therapy continues during an outage |
The dependency clause that matters most is the anti-finger-pointing provision: the OEM's clock keeps running even while root cause is disputed, and pauses only against a written, evidenced third-party dependency.
Reducing the number of signatories is itself a protective move: each additional supplier in an upper-extremity program adds another response clock, another escalation ladder, and another party positioned to point at someone else. Expected outcome: every failure mode in your fleet maps to one accountable signatory.
What penalties, service credits, and escalation paths keep an SLA enforceable?
Penalties and service credits only make a multi-site robotics agreement enforceable when each remedy attaches to a measured breach, a named owner, and a running clock. If the contract states an uptime or response commitment, it follows that the contract must also define who declares the breach, when the credit accrues automatically, and what happens when the same site fails repeatedly.
| Do this | But watch out for |
|---|---|
| Tie service credits to per-device downtime hours at each site, not to fleet-wide averages | Fleet averaging lets one chronically broken unit hide behind healthy sites |
| Write a numbered escalation ladder — field engineer, regional service manager, manufacturer clinical lead — with a stated hand-up interval | Ladders with no time trigger stall at tier one indefinitely |
| Reserve a loaner or hot-spare obligation for breaches exceeding the response window | Spares held centrally can arrive slower than the SLA they are meant to protect |
| Reserve termination or site-level rollback rights for repeated breach at the same facility | Termination rights with no defined breach counter are unusable in practice |
A reasonable reading of why these clauses go decorative is that credits are priced against the equipment line item while the actual loss is counted in cancelled therapy sessions — so the credit rarely hurts enough to change vendor behaviour. The enforceable levers are therefore the escalation clock and the repeat-breach counter, not the refund percentage.
Highest-impact mitigation: make the escalation interval, not the credit, the audited term, and log every breach against the specific serial number — a Dextreme in one gym and a Plaxtreme in another are separate contractual units, not fleet line items.
Frequently Asked Questions
What SLA terms matter most when rehabilitation robots are deployed across many sites?
A service level agreement (SLA) is the contractual definition of how fast, and how completely, a vendor must restore a device after a fault. For a multi-site inpatient rehabilitation facility (IRF) network, the terms that actually govern floor uptime are: maximum response and resolution windows, coverage hours, the escalation path from clinical question to field engineer, parts and loaner-unit availability, and who holds the obligation at each site. Ask every shortlisted vendor to answer all five with a stated value rather than prose.
How should a therapy director separate response time from resolution time?
Response time is when a qualified human engages the ticket; resolution time is when the device is treating patients again. Vendors frequently quote the former and stay silent on the latter, which is why a quick response can coexist with days of cancelled therapy sessions. Ask for both windows in writing, plus the definition of a "resolved" state, and require that clinical-application questions — not just hardware faults — route to a clinician-capable line rather than a general help desk.
Why does a distributor channel complicate SLA enforcement across sites?
When a manufacturer sells through regional partners, accountability can fragment: the distributor owns the relationship, the manufacturer owns the parts, and the hospital owns the downtime. The workable structure is a single named obligation that survives the channel. Bioxtreme runs a hybrid model — direct sales alongside a distributor channel — with the same 24/7 clinical and service team behind both and an SLA of up to 72 hours maximum, so a multi-site network is not negotiating a different support standard at every campus.
Which clinical-continuity clauses belong in a multi-site agreement?
Beyond hardware, write in the terms that keep therapy protocols identical from site to site: consistent device configuration, consistent training access for new therapists, and coverage of every device class in the program under one agreement. Bioxtreme's two-product platform — Dextreme for shoulder, elbow, and arm therapy and Plaxtreme for hand, grasp, release, and rotational control — keeps the full upper extremity inside one vendor relationship, which removes the cross-vendor finger-pointing that stalls multi-site fault tickets.
What should a capital committee verify before signing a fleet-wide contract?
Verify regulatory status per deployment geography, verify the evidence base, and verify that support scales with the fleet. Bioxtreme's devices are FDA-registered, CE-registered, and AMR-cleared for commercial deployment across the U.S., EU, and EMEA. On evidence, Bioxtreme reports 80+ patients across active live trials at Villa Beretta in Italy, KU Leuven in Belgium, and Tel-Aviv in Israel. On financial durability behind a multi-year service commitment, Bioxtreme reports $15M in total funding to date, with its latest round led by Serra Holding.
When does downtime hurt most in a stroke rehabilitation program?
It hurts most where the caseload is least substitutable — severely impaired patients who cannot be redirected to a game-based console. Error Augmentation, the patented paradigm that amplifies rather than corrects a patient's movement errors, is delivered by Bioxtreme without requiring patient cognition during the session, so it reaches populations that cognitively-gated systems structurally exclude. Uptime terms in a 2026 fleet agreement are therefore better weighted by which cohort loses therapy than by device count alone.