Jan Henrik Gundelsby

• Martini, Antonio; Nguyen, Phu; Ochoa, Lina; Gundelsby, Jan Henrik & Orucevic, Mili (2026). AI for Technical Debt Management: Summary from the AI4TD Workshop at XP2026. I Goldman, Alfredo; Marczak, Sabrina; Tonin, Graziela Simone; Eckstein, Jutta; Wang, Xiaofeng; Ferreira, Renato Cordeiro; Pinna, Andrea & Neumann, Michael (Red.), Agile Processes in Software Engineering and Extreme Programming – Workshops. Springer. ISSN 9783032392046. s. 91–99. doi: 10.1007/978-3-032-39205-3_10. Fulltekst i vitenarkiv Vis sammendrag

  Abstract Modern agile development and time-to-market pressures often lead to significant hidden technical debt (TD), which must be managed for sustainable software delivery. The growing use of artificial intelligence (AI) for code generation can further increase TD if applied improperly. At the same time, software processes are becoming increasingly data-driven. This creates an opportunity: using AI to enable data-driven TD management in agile environments. This workshop offers a platform for researchers and practitioners to share ideas, experiences, and future visions on applying AI for TD management in agile development. The workshop was held in a half day and it consisted of three phases informing each other: (1) scientific talks (i.e., two paper presentations and two keynotes) that framed current challenges and their relevance; (2) two industrial talks shaping these problems in practice; and interactive sessions (i.e., group discussions and a fish bowl panel) that allowed participants to reflect about the current status of AI with respect to TD and next steps. These discussions define a research agenda that spans measurement, practices, education, and legacy modernization, and call for tight collaboration between industry and academia to keep pace with AI-augmented software development.

• Gundelsby, Jan Henrik; Berntzen, Marthe; Grøneng, Camilla & Vigeland, Dan (2026). “I’ll Ask AI First”: How GenAI Disrupts Collaboration in Agile Teams. I Goldman, Alfredo; Marczak, Sabrina; Tonin, Graziela Simone; Eckstein, Jutta; Wang, Xiaofeng; Ferreira, Renato Cordeiro; Pinna, Andrea & Neumann, Michael (Red.), Agile Processes in Software Engineering and Extreme Programming – Workshops. Springer. ISSN 9783032392046. s. 61–70. doi: 10.1007/978-3-032-39205-3_7. Fulltekst i vitenarkiv Vis sammendrag

  Agile teams depend on collaboration, collective learning, and shared accountability to function effectively. As team members increasingly use generative AI (GenAI) tools, these foundational dynamics may be disrupted. In this study, we investigate how individual human–AI collaboration affects agile teamwork through thirteen interviews with software and technology professionals. Our findings show that GenAI reshapes team collaboration and learning while enabling individuals to work across traditional role boundaries, but also reinforces the continued necessity of human oversight and team accountability. We argue that these dynamics challenge the notion of agile teams, and call for updated models of agile teamwork.

• Irgens, Marius; Orucevic, Mili; Boye, Mads; Gundelsby, Jan Henrik & Martini, Antonio (2025). DebtQuest: Discover Technical Debt Management Issues with Survey Visualization. IEEE Working Conference on Software Visualization (VISSOFT). ISSN 2379-7576. 2025, s. 25–35. doi: 10.1109/vissoft67405.2025.00011. Fulltekst i vitenarkiv Vis sammendrag

  While Technical Debt (TD) manifest in systems as code-related symptoms, it is often caused by management-related issues. However, existing TD visualization tools are primarily code-based, and limited in addressing organizational factors. In this context, we introduce DebtQuest: an interactive, survey-based visualization tool designed to help assess an organization’s TD management practices. DebtQuest enables stakeholders to explore insights, pinpoint issues, perform benchmarking, and compare changes across time. Complementing the capabilities of code-based tools to discover TD symptoms, DebtQuest supports the discovery of underlying organizational causes. DebtQuest’s design is grounded in task analysis and a targeted review of relevant visualization techniques. It was refined through real-world use and evaluated by industry stakeholders. The contribution of this work is an accessible and flexible visualization design that extends recent research in TD management, revealing organizational insights that are difficult to discover using code-based tools.

• Ulfsnes, Rasmus; Moe, Nils Brede; Emmerhoff, Jostein; Floryan, Marcin; Griva, Anastasia & Gundelsby, Jan Henrik [Vis alle 8 forfattere av denne artikkelen] (2024). Responsible AI in Agile Software Engineering - An Industry Perspective. I Marchesi, Lodovica; Goldman, Alfredo; Lunesu, Maria Ilaria; Przybyłek, Adam; Aguiar, Ademar; Morgan, Lorraine; Wang, Xiaofeng & Pinna, Andrea (Red.), Agile Processes in Software Engineering and Extreme Programming – Workshops: XP 2024 Workshops, Bozen-Bolzano, Italy, June 4–7, 2024, Revised Selected Papers. Springer. ISSN 9783031727818. s. 33–41. doi: 10.1007/978-3-031-72781-8_4. Fulltekst i vitenarkiv
• Stray, Viktoria; Gundelsby, Jan Henrik; Ulfsnes, Rasmus & Moe, Nils Brede (2022). How agile teams make Objectives and Key Results (OKRs) work. I Carleton, Anita (Red.), ICSSP'22: Proceedings of the International Conference on Software and System Processes and International Conference on Global Software Engineering. Association for Computing Machinery (ACM). ISSN 9781450396745. s. 104–209. doi: 10.1145/3529320.3529332. Fulltekst i vitenarkiv Vis sammendrag

  Today, many companies allow their employees to work from anywhere, which has changed how employees coordinate their work and align toward the same goals. Objectives and Key Results (OKRs) is a goal-setting framework applied in such distributed settings. This research aimed to investigate how OKRs are used in large-scale agile contexts. We interviewed team members and analyzed documents, including a survey. Our study’s results provide both enabling and limiting situations that make team members’ utilization of the framework either easier or more difficult. We found that OKRs aided knowledge sharing and improved transparency between teams. We present four strategies used for overcoming challenges and maximizing the benefits of using a goal-setting framework. An important takeaway is that companies that employ OKRs must support their employees, especially in defining key outcomes that align and encourage teams toward a common goal.

• Gundelsby, Jan Henrik (2018). Enabling autonomous teams in large-scale agile through architectural principles. I Wagner, Stefan; Hoda, Rashina & Aguiar, Ademar (Red.), Companion Proceedings: 19th International Conference on Agile Software Development (XP2018). Association for Computing Machinery (ACM). ISSN 9781450364225. doi: 10.1145/3234152.3234183. Fulltekst i vitenarkiv Vis sammendrag

  Teams in large-scale projects and programs need to reach agreement on many decisions with experts, managers and other stakeholders. The need for aligning the work and the process with the rest of the organization reduces team autonomy. Coordination by architecture is one strategy to handle this challenge. Consequently, we did a case study of a large-scale software program consisting of nine teams, to understand how the architecture can enable team autonomy. Initially teams had limited autonomy, because of high task dependencies with other teams and experts. By introducing an architecture based on business domains and APIs, teams got full responsibility for a set of components, and solved the alignment problem by letting other teams access the resources through an API. The new architectural strategy can be understood as structuring by business domains and APIs, instead of features that span the whole code base.

Se alle arbeider i NVA

• Moe, Nils Brede & Gundelsby, Jan Henrik (2026). Norsk forskning - Hva gjør norske organisasjoner med agentisk utvikling - Jan Henrik Gundelsby og Nils Brede Moe. Fulltekst i vitenarkiv Vis sammendrag

  This is "Norsk forskning - Hva gjør norske organisasjoner med agentisk utvikling - Jan Henrik Gundelsby og Nils Brede Moe

• Moe, Nils Brede; Gundelsby, Jan Henrik & Reistad, Silje Rognsvåg (2026). – De som virkelig får til KI, jobber først tregere. [Journal]. kode24.no. Fulltekst i vitenarkiv Vis sammendrag

  Utviklere må få tid til å lære seg teknologien skikkelig, sammen, konkluderer Sintef og Knowit etter lang forskning.

• Gundelsby, Jan Henrik (2024). Norsk forskning. Fulltekst i vitenarkiv
• Conboy, Kieran; Moe, Nils Brede; Stray, Viktoria & Gundelsby, Jan Henrik (2023). The Future of Hybrid Software Development: Challenging Current Assumptions. IEEE Software. ISSN 0740-7459. 40(2), s. 26–33. doi: 10.1109/MS.2022.3230449. Fulltekst i vitenarkiv
• Hanssen, Geir Kjetil; Sporsem, Tor Thorsrud & Gundelsby, Jan Henrik (2022). Datadrevne bedrifter gjør lurt i å lytte til magefølelsen også. Fulltekst i vitenarkiv
• Gundelsby, Jan Henrik (2022). Organisering av produktutvikling i store organisasjoner. Fulltekst i vitenarkiv
• Gundelsby, Jan Henrik (2022). How agile teams make Objectives and Key Results (OKRs) work. Fulltekst i vitenarkiv
• Gundelsby, Jan Henrik (2021). Forskningsprosjektet Transformit. Fulltekst i vitenarkiv
• Sporsem, Tor Thorsrud & Gundelsby, Jan Henrik (2021). Forskningsprosjekt på datadrevet digital transformasjon - hvordan vi forsker og tidlige resultater. Fulltekst i vitenarkiv
• Gundelsby, Jan Henrik & Johnsen, Steinar Henrik (2002). Evaluation of the EJB 2.0 and COM+ component models using a common testcase. Universitetet i Oslo. Fulltekst i vitenarkiv Vis sammendrag

  A component model is a defined set of services that assist the developer with traditionally difficult tasks such as transaction handling, synchronization, and security. The two component models Enterprise Java Beans 2.0 and COM+ are the two main competitors on the current market, and they are compared both from a service and performance perspective. To benchmark performance, implementations of a common test case were conducted in both technologies. WebLogic 6.1 was used for the Enterprise Java Beans implementation. The set of services offered by the component models are compared systematically, and mapped according to the authors’ own experience and to the test implementation. The primary goal of this thesis is to provide an unbiased comparison of the two component models. The findings of this thesis indicate that the two component models are quite analogous with respect to services. From the performance point of view, COM+ proved to be somewhat faster than Enterprise Java Beans.

Se alle arbeider i NVA