Governing the Integration of AI-Assisted and Semi-Autonomous Tools in Enterprise Software Development: A Qualitative Study
(2026) INFM12 20261Department of Informatics
- Abstract
- AI-Assisted and Semi-Autonomous tools are becoming relevant for enterprises. The adoption of these tools requires understanding how different practitioners perceive and use them responsibly in their organizations. This thesis examines how enterprise software development practitioners govern AI-assisted and semi-autonomous tools in everyday development workflows. As these tools increasingly support different Software Development Lifecycle processes like coding, testing, debugging, documentation and reviewing, organizations must maintain control, accountability, and responsibility within shared workflows between same or different roles. Based on a qualitative interpretive study with ten software development practitioners, the study uses... (More)
- AI-Assisted and Semi-Autonomous tools are becoming relevant for enterprises. The adoption of these tools requires understanding how different practitioners perceive and use them responsibly in their organizations. This thesis examines how enterprise software development practitioners govern AI-assisted and semi-autonomous tools in everyday development workflows. As these tools increasingly support different Software Development Lifecycle processes like coding, testing, debugging, documentation and reviewing, organizations must maintain control, accountability, and responsibility within shared workflows between same or different roles. Based on a qualitative interpretive study with ten software development practitioners, the study uses abductive thematic analysis to examine governance as a sociotechnical phenomenon. The findings show that practitioners govern AI integration through three interdependent practices: governed oversight of AI use, bounded AI autonomy in development work, and distributed responsibility for AI-supported outcomes. AI-assisted tools are mainly treated as productivity and support mechanisms, whereas semi-autonomous tools require stronger boundaries because they can execute delegated tasks with less direct human input. Practitioners rely on review gates, validation routines, traceability, prompting discipline, access restrictions, and human approval to keep AI contributions accountable. The thesis contributes to Information Systems research by conceptualizing the autonomy-oversight-responsibility loop as a dynamic mechanism for responsible AI
integration. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/student-papers/record/9240243
- author
- Kokkalis, Konstantinos LU and Atul Shastri, Aditya LU
- supervisor
- organization
- course
- INFM12 20261
- year
- 2026
- type
- H1 - Master's Degree (One Year)
- subject
- keywords
- Responsible AI integration, AI-assisted, software development, semi-autonomous, AI tools, enterprise, human-AI collaboration, autonomy, oversight, responsibility, socio-technical systems
- language
- English
- id
- 9240243
- date added to LUP
- 2026-06-18 09:12:07
- date last changed
- 2026-06-18 09:12:07
@misc{9240243,
abstract = {{AI-Assisted and Semi-Autonomous tools are becoming relevant for enterprises. The adoption of these tools requires understanding how different practitioners perceive and use them responsibly in their organizations. This thesis examines how enterprise software development practitioners govern AI-assisted and semi-autonomous tools in everyday development workflows. As these tools increasingly support different Software Development Lifecycle processes like coding, testing, debugging, documentation and reviewing, organizations must maintain control, accountability, and responsibility within shared workflows between same or different roles. Based on a qualitative interpretive study with ten software development practitioners, the study uses abductive thematic analysis to examine governance as a sociotechnical phenomenon. The findings show that practitioners govern AI integration through three interdependent practices: governed oversight of AI use, bounded AI autonomy in development work, and distributed responsibility for AI-supported outcomes. AI-assisted tools are mainly treated as productivity and support mechanisms, whereas semi-autonomous tools require stronger boundaries because they can execute delegated tasks with less direct human input. Practitioners rely on review gates, validation routines, traceability, prompting discipline, access restrictions, and human approval to keep AI contributions accountable. The thesis contributes to Information Systems research by conceptualizing the autonomy-oversight-responsibility loop as a dynamic mechanism for responsible AI
integration.}},
author = {{Kokkalis, Konstantinos and Atul Shastri, Aditya}},
language = {{eng}},
note = {{Student Paper}},
title = {{Governing the Integration of AI-Assisted and Semi-Autonomous Tools in Enterprise Software Development: A Qualitative Study}},
year = {{2026}},
}