Description
Framna is seeking a final-year master’s student to work on a thesis project in its Stockholm studio, starting in January 2027. The project will define a benchmark of realistic mobile-app user tasks and evaluate how reliably AI agents can complete them through accessibility-tree navigation, vision-based screenshot control, and declared capability interfaces such as App Intents or App Actions. The student will measure task completion, steps, latency, token cost, recovery, and failure modes, identify factors affecting agent operability, and recommend which actions agents should perform autonomously versus with user confirmation.
