Blind and severely visually impaired people often rely on routines and memory to locate everyday objects. When keys, medication, documents, or assistive devices are misplaced, searching can cause stress, time loss, and increased dependence on others. Existing assistive technologies mainly support navigation, immediate object recognition, or physical tracking and may require cameras, cloud services, specialised hardware, or attached tags. Lightweight systems that store and retrieve object-location information over time remain underexplored.

The thesis combined a literature review, requirements engineering, and input from a blind accessibility expert. The findings were translated into functional and non-functional requirements for accessibility, usability, privacy, and reliability. Based on these requirements, a modular prototype was implemented using local speech recognition, language processing, SQLite storage, and speech synthesis. The system was evaluated with 12 participants across 16 task-based scenarios.

The evaluation confirmed the technical feasibility of WhereIsIt. Most test cases were completed successfully, with basic recording, storage, retrieval, and error-handling tasks performing particularly well. More complex operations and fully non-visual use were less reliable, highlighting the need for clearer feedback, confirmation, and error recovery. Overall, the prototype demonstrates the potential of a lightweight, privacy-conscious, and locally operated assistive system, although further testing with blind users is required.