Careers

Why the hiring bar sits where it does

Everything described elsewhere on this site — prediction, sensors, scored inspections — produces signals. None of it cleans a room, and none of it notices the thing that was not being measured.

A sensor reports that a restroom took four hundred visits. A person walking past notices that a fixture is loose, that a door is not closing properly, that a room smells wrong in a way no reading captures. The proactive model’s most valuable inputs still come from someone deciding that something is worth mentioning.

That is a hiring problem before it is a technology problem, and it is why the selection process is stricter than the industry norm. There is no training module for caring whether the room is right.

Selection

A structured interview, reference checks, and background verification appropriate to the environments the role covers. Medical and government sites frequently carry their own additional requirements, handled per account.

We are deliberately slower here than competitors who staff by volume. A contract covered by people who were hired quickly and will leave quickly cannot deliver a standard that depends on familiarity with a specific building.

Onboarding, then site induction

Two layers, because they answer different questions.

General onboarding covers chemical handling and incompatibilities, PPE, equipment, colour-coded segregation and the procedures that apply everywhere. Detail on Safety.

Site induction happens before an account starts and covers what is true only there: hazards, restricted areas, access rules, evacuation routes, chemical restrictions, and the protocol for spaces with their own requirements. It is repeated when a site materially changes, because the only useful induction is the current one.

Tools that assume the actual conditions

Most operational software in this industry is designed as though the person using it is at a desk. The result is predictable — entries typed at the end of a shift, from memory, in a batch, with timestamps that are fiction.

CleanPilot is a voice assistant instead: clock in by saying so, create a task by describing it, ask what is next. It works in English, Portuguese and Spanish, and the highest-frequency actions run on deterministic shortcuts rather than model interpretation, because a shift record is a payroll record.

This is a working condition as much as an engineering decision. Being expected to type on a phone with wet gloves, and then being judged on the quality of what you entered, is a bad job made worse by software.

Work that is visible

In a conventional cleaning operation, careful work leaves no trace and only failures get attributed. Someone who quietly keeps a difficult floor in good condition for a year is indistinguishable, on paper, from someone who does the minimum.

Here, completion records, inspection scores and client feedback attach to the people who did the work. That cuts both ways and is meant to — but the side that goes unmentioned in most quality-control pitches is that it finally makes good work legible.

It also feeds Protex Verified™, which is signed by both a supervisor and the client. Certification the team earns rather than a rating the company awards itself.

Getting in touch

Openings depend on which accounts are starting and where, so they vary through the year rather than sitting posted continuously. Reach us through the contact page or by phone during business hours, and say what kind of role and which area you are interested in.