This peer-reviewed study evaluated whether a soft robotic hand orthosis (RELab tenoexo 2.0) could feasibly support functional hand rehabilitation in chronic stroke patients across clinic and home settings. The prospective, single-arm feasibility design was conducted at Tan Tock Seng Hospital, Singapore from November 2023 to January 2025, enrolling 8 participants who were >6 months post-stroke. The participants went through a 11-week protocol utilizing device-logged usage data, clinical assessments (ARAT, FMA-UE), and usability questionnaires. The soft robotic orthosis had a spring-blade actuation mechanism and was not AI-based. It had no direct comparator group. Data collected from the study showed 81% compliance to suggested usage time and 150.8±82.9 grasps per hour at home. Clinical outcomes demonstrated that 6/8 participants achieved meaningful FMA-UE improvements (≥5.75 points) and 3/8 achieved meaningful ARAT improvements (≥5.7 points), retained at one-month follow-up. However, there are significant limitations which compromise the analysis of the results, like selection bias. This occurred because occupational therapists were not blind and had previously treated participants. There were also technical reliability issues which required 0-5 home repair visits per participant due to device failures. Finally, there was poor completion of at-home logs, which limited usage insights. There were no confidence intervals provided for any outcomes, only standard deviations. No external validation was performed, and comprehensive fairness analysis is missing despite Singapore's diversity—no demographic data on race, ethnicity, socioeconomic status, or education were reported. At the bedside, this evidence demonstrates only that robotic hand training is potentially feasible for some chronic stroke patients, but provides no evidence this approach improves outcomes compared to standard rehabilitation. The single-arm design, frequent device failures, and substantial technical support requirements make clinical implementation impractical without significant design improvements and larger controlled trials directly comparing robotic training to standard care before any clinical efficacy can be established.