• A FlyTrex remote pilot in command supervises deliveries from an operations center rather than flying the drones, per MIT Technology Review.
• One pilot may oversee up to 10 drones at once at peak times; the profile states this as a peak, not a formal ratio.
• Collision avoidance relies on unmanned aircraft systems traffic management, a network where drones share positions in real time.
• Weather is named as the most common hazard, but no intervention, abort or incident counts are given.
• Inside Unmanned Systems reports the FAA's BVLOS proposal would require a designated operations supervisor and flight coordinator, neither needing a formal FAA pilot certificate.

Delivery drones raise a staffing question as well as an aircraft question. MIT Technology Review profiles Trevor Wischnewsky, a remote pilot in command at FlyTrex, who watches the drones rather than flying them and may oversee up to 10 at once at peak times. That figure is the only evidence the profile offers for the staffing model, and it comes with no failure data.

The human is a safety net, not a pilot

The role is supervision. Trevor Wischnewsky, a remote pilot in command at FlyTrex, monitors deliveries from an operations center using GPS, lasers and sensors rather than flying the aircraft, MIT Technology Review reports. FlyTrex's hexacopters deliver within a three-mile radius. The company requires an FAA drone pilot certification and a weeks-long FAA-approved course so that a pilot can take control of a drone when needed. In his words, he is there as a safety net.

Traffic management does the deconflicting

Collision avoidance sits in the network, not with the pilot. MIT Technology Review describes the Dallas–Fort Worth area as a testing ground for unmanned aircraft systems traffic management, a network where drones share positions in real time, including beyond a pilot's line of sight. Wischnewsky says that once a drone is cleared to fly, the system takes care of the rest. Zipline, Amazon and Alphabet's Wing also fly over Dallas, so the arrangement is shared infrastructure rather than one company's stack.

Weather is the hazard the profile names

The profile identifies weather as the most common hazard. MIT Technology Review describes one day when a sudden gust hit a drone shortly after takeoff; Wischnewsky intervened, returned it safely and saved an ice cream cake from being dropped. That is one anecdote, not a rate. The profile gives no count of interventions, aborts or incidents, so how often the safety net is actually used cannot be estimated from it.

The proposed federal rule splits compliance from flight oversight

Inside Unmanned Systems reports that the FAA's beyond-visual-line-of-sight proposal would require each certificate holder to designate an operations supervisor, with a flight coordinator overseeing individual flights, and that neither position would require a formal FAA-issued pilot certificate. That report does not say how many aircraft one person may oversee, and this piece makes no claim about a final rule. FlyTrex's certification requirement is therefore a company practice, not something that report shows the proposal demands.

Verdict: the pattern works on paper, the ratio is unproven

The supervise-and-intervene pattern looks workable, but the 10-to-1 figure should not be treated as a benchmark. It is a peak reported in one profile, with no intervention rate, abort count or incident log behind it. Teams planning similar human-on-the-loop systems should ask for those three numbers before sizing a supervision team, and should expect weather to set the real limit. Until the numbers exist, 10 drones per pilot is an operator's anecdote.