Salamanca schools plan a classroom pilot with humanoid assistant Sally

Salamanca City Central School District says it will introduce a stationary humanoid robot named Sally in a limited classroom pilot during the 2026-27 school year. The western New York district plans to begin with a small cohort in a Woz Ed STEAM program rather than deploy the system across every classroom.

The device comes from Realbotix, which describes it as an AI-powered teaching assistant. The district's own announcement is more specific about the operational boundaries: classroom teachers will remain in control, the curriculum will be approved locally, and Sally is intended to support lessons rather than replace educators. Claims that this is the first such US deployment come from the vendor and should be treated as positioning, not an independently established national record.

According to Salamanca's announcement, the pilot is designed with several restrictions:

• •Sally will remain stationary rather than move freely around the classroom.
• •It will not browse or search the open internet during lessons.
• •The district says it will not collect personally identifiable student information.
• •The device is not supposed to record student audio or video.
• •The district says classroom data will not be transmitted to Realbotix.
• •Teachers will direct its use and retain responsibility for instruction.

These are meaningful design constraints, but they are not yet independently audited outcomes. Even without names or recordings, any student-code or personalization mechanism still needs clear rules for retention, access, deletion and correction. The district has not yet published evidence from classroom use showing how the system behaves when it misunderstands a student, produces an inaccurate answer or encounters a situation outside its approved material.

Independent reporting from New York Focus and the Olean Times Herald confirms the local pilot and adds procurement and community context. It does not provide evidence that learning outcomes have improved, because the classroom trial has not begun.

The responsible way to evaluate Sally is as a bounded education-technology pilot, not as proof that humanoid teachers have arrived. A useful evaluation should compare learning and workload outcomes against ordinary classroom tools, document failure and escalation paths, and let families understand what data is created and how it can be removed.

The district's restrictions reduce several obvious risks, especially uncontrolled web access and undisclosed recording. The next standard is evidence: transparent pilot metrics, incident reporting, teacher feedback and a clear decision process before any expansion. Without those results, the most defensible conclusion is simply that Salamanca is testing a novel interface under stated safeguards.