HumanoidUptime
Last evidence review · 04 Sep 2026
← All evidence records

HU-R007 / Deployment evidence

Unnamed Sanctuary GPR × Canadian Tire / Mark's

Deployment / test evidenceManufacturer-reported pilot
Reliability evidenceManufacturer-reported task count; no reliability rate

Sanctuary reports a one-week retail pilot and 110 correctly completed tasks, without publishing the attempts, operating time or control mode needed to assess reliability.

Multi-task retail pilotMark's store, Langley, British Columbia, CanadaCompleted one-week pilotRecord first published 2026-09-04Last record change 2026-09-04Evidence current to 2026-09-04
1 wkretail pilot
110tasks reported correct
1named store
Cite this recordHU-R007 · rev. 7 · Public Data v1.4.0

HumanoidUptime. “Unnamed Sanctuary GPR × Canadian Tire / Mark's” (HU-R007), rev. 7, evidence current to 2026-09-04. HumanoidUptime Public Data v1.4.0 (2026). https://humanoiduptime.com/id/HU-R007

Download CSL JSON →
Persistent identity
HU-R007 · revision 7
Evidence cutoff
2026-09-04
Dataset release
v1.4.0 · snapshot 2026-09-04
Immutable dataset
Versioned JSON
JSON SHA-256
0e174ef9f5d6de3f140478daa628032f4610761ba3794ec124bd310b33ec9705

What this means

The short version, without reliability jargon

What we know

Sanctuary says one general-purpose robot spent a week performing retail tasks at a Canadian Tire-owned Mark's store and completed 110 tasks correctly.

What remains unknown

The source does not identify the robot version, total attempts, failed tasks, operating hours or how much of the work was autonomous rather than remotely controlled or assisted.

Why it matters

The pilot shows that a robot performed varied tasks in a real store, but it does not reveal its success rate or operational reliability.

Evidence verdict

Established. Claimed. Unknown.

EstablishedOperator or independent evidence

No operator-corroborated or independent finding in the reviewed sources.

ClaimedManufacturer or capability claim
  • Deployment scopeOne-week pilot at a Canadian Tire-owned Mark's retail store in Langley[1] [2]
  • Task breadth110 retail-related tasks completed correctly[1]
  • Task successTasks described as completed correctly[1]
  • Work settingFront- and back-of-store activities in a real-life retail environment[1]
  • Robot identityUnnamed Sanctuary general-purpose robot[1] [4]
  • Operator responseCanadian Tire executive describes operational-efficiency exploration and shifting people toward higher-value work[1] [2]
Still unknownNo qualifying public value
  • Robot model and hardware configuration
  • Task attempts, retries and success-rate denominator
  • Autonomous, teleoperated and human-assisted shares
  • Runtime, downtime, interventions and recovery data

What the evidence says

Sanctuary establishes that a general-purpose robot worked for one week at a Canadian Tire-owned Mark's store and says 110 retail-related tasks were completed correctly. The source does not identify the robot generation, distinguish unique task types from executions, publish failed attempts, or state whether the work was autonomous or teleoperated.

01 / Reliability fields

What is known—and missing

Not disclosed
Not disclosed
Not disclosed
Not disclosed
Not disclosed
110 tasks reported correct
Energy / charging
Not disclosed
Pilot mode not disclosed
1 robot described
Independently measured Operator corroborated Manufacturer reported Capability claim Not disclosed Not applicable to this record

02 / Claims and evidence

What each source supports

Deployment scope Manufacturer reported

One-week pilot at a Canadian Tire-owned Mark's retail store in Langley

Sanctuary names the operator, banner, city and duration. Canadian Tire's own venture page confirms its investment relationship with Sanctuary but does not independently describe the pilot.

Sources: [1] [2]
Task breadth Manufacturer reported

110 retail-related tasks completed correctly

Examples include picking and packing, cleaning, tagging, labelling and folding. The source does not clarify whether 110 means distinct task types, executions or another counting unit.

Sources: [1]
Task success Manufacturer reported

Tasks described as completed correctly

No total attempt count, failed-task count, retry rule or success criterion is published, so a task-success rate cannot be calculated.

Sources: [1]
Work setting Manufacturer reported

Front- and back-of-store activities in a real-life retail environment

This establishes a customer commercial facility rather than a laboratory. It does not establish continuous production use, paid output or operation during normal customer traffic.

Sources: [1]
Autonomy Not disclosed

Not disclosed for the Mark's pilot

Sanctuary later explained that its robots can work autonomously while most videos in its Robots Doing Stuff series are . That general disclosure does not determine the mode used in this pilot.

Sources: [1] [3]
Robot identity Manufacturer reported

Unnamed Sanctuary general-purpose robot

The deployment release does not name a generation. Sanctuary says Phoenix was unveiled on 16 May 2023, after the January pilot, so HumanoidUptime does not retroactively label the tested system as Phoenix.

Sources: [1] [4]
Operator response Manufacturer reported

Canadian Tire executive describes operational-efficiency exploration and shifting people toward higher-value work

The quotation is attributed to a named Canadian Tire executive but appears in Sanctuary's release. No standalone Canadian Tire result or quantitative efficiency evidence was found in the reviewed source set.

Sources: [1] [2]

03 / Deployment timeline

What happened, and when

  1. Retail pilot conducted

    Sanctuary says a general-purpose robot spent one week at the Mark's store in Langley and attempted retail work.

    [1]

  2. Sanctuary publishes deployment account

    The manufacturer reports the site, duration, task examples and 110 tasks completed correctly.

    [1]

  3. Phoenix unveiled later

    Sanctuary's subsequent product chronology dates the public Phoenix unveiling after the retail pilot.

    [4]

  4. Sanctuary explains teleoperation practice

    A general methodology post confirms that teleoperation is widely used for demonstrations and training but does not identify the mode of the Mark's pilot.

    [3]

  5. HumanoidUptime evidence audit

    Record opened with the pilot facts preserved separately from unknown autonomy, reliability and success-rate denominators.

    [1] [2] [3] [4]

04 / Sources

Trace every finding

  1. Source 1Manufacturer statement

    Sanctuary AI Deploys First Humanoid General-Purpose Robot Commercially ↗

    Sanctuary AI · Published 2023-03-07 · Accessed 2026-09-04

  2. Source 2Operator investment relationship

    Roller Labs Ventures portfolio: Sanctuary ↗

    Canadian Tire Corporation · Published date not stated · Accessed 2026-09-04

  3. Source 3Manufacturer methodology

    AI Robotics and the Case for Teleoperation ↗

    Sanctuary AI · Published 2024-01-24 · Accessed 2026-09-04

  4. Source 4Manufacturer product chronology

    A Retrospective Look: Phoenix and Carbon Unveiling ↗

    Sanctuary AI · Published 2023-08-10 · Accessed 2026-09-04

Evidence current to2026-09-04

Found something newer or more precise?

Submit evidence