The platform · our software
The devices listen.
The platform understands.
This is the software half of Dr. Elevage — the AI engine that reads every trip, the console your team opens each morning, and the growing operating record behind every assessment. The magnetic hardware can be removed and replaced quickly; the operating history cannot.
01 · Division of labor
Hardware collects the data. Software turns it into answers.
The sensors record every trip. The value is in reading it: what a given signal means on this particular 18-year-old machine, in this shaft, under this duty cycle — and having the evidence to show for it afterwards. That understanding comes from years of data, and it lives in the software.
Collect
- Vibration on 6 axes, acoustics across the audible band, speed curves, door-zone behavior
- Every trip, on every unit, recorded by the device at the moment it happens
- Magnetic, non-invasive, removable without tools
Understand, decide, prove
- Learns a health baseline for each individual elevator and scores every trip against it
- Identifies the suspected component, travel segment and risk grade before failure
- Ranks the whole fleet so the technician’s day is planned by measured risk
- Keeps the record: what it looked like before the repair, and after
- Uses verified repair outcomes as feedback to help refine future assessments
You are not buying an accelerometer. You are buying answers to questions you could not answer before — and a record you can hand to anyone who asks.
02 · The console
Four screens, in the order you would actually use them.
Click through the panels below. This is the shape of the working day: open the fleet view, act on what is flagged, verify the fix, hand over the record. The same four screens run in a browser at the office and on a phone in the shaft.
Console view with real deployment data. Click to enlarge.
Open "attention required" and you get the list ranked by measured risk; any unit is one click from its speed curves, 3-axis vibration and acoustic waveforms. A fleet list shows what you have; a risk-ranked list helps show where to look first.
A "start-up jitter" alert, opened: two trips from the lowest floor, raw 3-axis trace, abnormal window already boxed.
Real console screens. Click either image to enlarge.
An alert that does not tell a technician where to stand is just noise with a serial number. Every alert carries the component, the location and the recommended inspection point — which is also what makes it dispatchable as a work order with far less back-and-forth.
- Vertical scale
- ±1.7
- Abnormal phases
- 3 of 3
- Vertical scale
- ±0.8
- Abnormal phases
- 0 of 3
Mobile app view — the traces are the recorded data. Grey = Z axis (vertical), red = smoothed trend. Note the vertical scale: ±1.7 before, ±0.8 after.
The delta is the deliverable. A customer cannot audit whether a rope was replaced properly, but anyone can see a whole-trip trace go from violent to smooth on the same run — and the same record holds up if the outcome is ever disputed.
A repair diagnosis report generated by the platform — the compensation-rope case from the previous screen. Unit identifier partly masked.
Compliance evidence pack
Useful in inspection conversations: unit identity, health status, risk grade, trend history, complete service log.
Owner / board report
Plain-language: which units need money this quarter, which do not, and what the last round of spending actually achieved.
Underwriting file
Consistent, comparable condition-risk grades for individual elevators, supported by the underlying operating data.
Incident data log
Complete, passively collected record before, during and after an event — for investigation rather than argument.
03 · Inside the AI engine
What the engine does with every trip, on every unit.
This runs continuously in the background — not on demand, not when someone remembers to check. Four stages, and the fourth one is what makes the first three get better over time.
Condition the data
Each trip is tagged with load, direction, speed and floor range, so a full car going up is only ever compared with other full cars going up. Most false alarms in this category come from skipping this step.
Score against its own baseline
Not a fixed industry threshold — this unit’s own recorded normal. That is why an 18-year-old machine can be monitored usefully, without raising an alert every single day.
Match, name and locate
Sustained drift is matched against a library of 50+ fault signatures built from real-world elevator failures. The output is a component, a floor range, a severity grade and a recommended inspection point.
Learn from the verified outcome
The technician confirms on site and the result — whether the initial assessment was right or wrong — is fed back into the system. Over time, these verified outcomes help refine future assessments on your fleet.
04 · The data asset
The record is the thing you are actually buying.
A sensor tells you about today. A multi-year, per-unit, passively collected record tells you what a building is worth, which vendor is doing the work, whether the last repair held, and what to budget for next year. That record is not a by-product of the product — it is the product.
It compounds
Month one, the platform knows what this elevator sounds like. Month twelve, it knows what it sounded like before the last three repairs, how quickly it degrades in summer, and which of your buildings works its elevators hardest. None of that can be bought in — it can only be accumulated.
A separate operating record
An OEM platform is kept by the company that built the elevator. A maintenance log is kept by the company being assessed. This record is maintained separately from routine maintenance records, giving third parties a more neutral basis for review.
It travels with the asset
Change maintenance vendor and the history stays with the building, not with the contractor who is leaving. For an owner, that alone changes the balance of a renewal negotiation.
It is comparable
Because every unit is measured the same way with the same method, two elevators of different brands in different cities can sit in the same ranked list. That is what makes portfolio-level decisions — and underwriting — possible at all.
Every month it runs, the platform gets harder to replace — not because we lock you in, but because nobody else has your elevators' history.
05 · Reports & who reads them
One record. Four audiences. One click.
Every finding can be traced back to the underlying data. Health status, trends and effectiveness checks are built from automatically captured operating records, creating a clear trail from anomaly to action to verification.
| Section | What it contains | Source | Who reads it |
|---|---|---|---|
| Unit identity & configuration | Registration number, brand, model, year, rated speed and load, building and shaft position | Set at commissioning | Everyone |
| Health status & risk grade | Current score, grade, and the specific deviations driving it | AI engine | Owner, regulator, insurer |
| Trend history | Vibration and acoustic trend over the reporting period, with alert events marked in place | Continuous sensing | Owner, insurer |
| Maintenance record | Every service visit detected passively, with the date, window and duration — not a signed sheet | Passive detection | Owner, regulator |
| Effectiveness validation | Before/after comparison at identical measurement points for each intervention | Paired measurement | Owner, maintenance company |
| Open findings & priority | What is still outstanding, ranked, with the recommended inspection point for each | AI engine | Maintenance company |
06 · Access
Who sees what.
The same underlying record, surfaced differently for each role — so everyone can review the same traceable measurements rather than work from separate versions of the story.
Maintenance company
Full fleet console, alerts, dispatch context, before/after tooling, and the reports they issue to their own customers.
Building owner / property
Their own buildings only: health, risk, budget prioritization and the verified outcome of every visit they paid for.
Inspection firm
Standardized measurement procedure, exportable curves, and defensible evidence attached to each finding they sign.
Insurer / regulator
Comparable per-unit risk grades across a portfolio, plus the complete data log when an incident needs investigating.
Dr. Elevage provides a consistent condition-data layer across different stakeholders — maintenance companies, owners, inspection firms and risk teams. Because the measurements are collected independently of the elevator controller, the same underlying data can support review across these different roles.
07 · Common questions
What the record can and cannot do.
Can the operating record be edited after the fact?
Not by the party being assessed. Operating records are written by the device and uploaded as recorded, and the console gives no one an interface to edit them. There is no manual check-in, no uploaded photo and no technician-entered field anywhere in that chain.
Is anything entered by hand at all?
One thing, once. The unit's nameplate record — registration number, brand, model, year, rated speed and load, building and shaft position, floor map — is entered when the unit is commissioned. Everything after that is captured automatically, and the nameplate record cannot be changed by the maintenance contractor afterwards.
Does customer feedback change the record itself?
No. Two things are kept apart. Verified repair outcomes can be used as feedback to help refine future assessments. The operating record that gets quoted to an owner, an inspector or an insurer — before/after data, detected service windows — is kept separate from that improvement process and is not editable.
Does this replace statutory inspection or the maintenance log?
No. It supplements them. It does not replace statutory periodic inspection, servicing by a qualified maintenance contractor, or any record either of those requires. What it adds is one set of numbers that nobody typed in, which every party can quote from.