What the Jenkins Logs Won’t Tell You: Using an AI Agent to Capture the Lost ‘Bank Memory’ Behind a 76% Sprint Velocity Gain and Whether Another Community Bank Can Borrow It Without the Original Team
DOI:
https://doi.org/10.63282/3050-9262.IJAIDSML-V5I3P126Keywords:
Lost Bank Memory, LLM Agents, Knowledge Transfer, Community Banks, Fintech Replicability, CI/CD Tacit Knowledge, Sprint Velocity, Retrieval-Augmented Generation, Agile Institutional MemoryAbstract
Community banks are increasingly adopting enterprise-scale FinTech backbones, but replicating a success story like a 76% jump in sprint velocity is rarely as simple as copying code or rerunning Jenkins pipelines. In this study, we argue that much of what made that velocity gain possible was never written down it lived in undocumented API workarounds, late-night CI/CD fixes, team-specific Scrum rituals, and implicit AWS configuration choices. We call this hidden asset lost bank memory. To explore whether a second, resource-constrained community bank could borrow this success without the original engineering team, we design an LLM-based agentic framework that mines not only structured artifacts (Git logs, JIRA histories, Jenkins traces) but also semi-structured conversational data (Slack threads, code review comments, incident postmortems). Our agent retrieves, infers, and contextualizes the tacit decisions behind the original implementation. Through a simulated transfer experiment across two different community bank settings, we find that raw replication fails in 63% of key workflows, but our AI agent reduces missing-knowledge failures by nearly half. The paper concludes that lost bank memory is a measurable, recoverable asset but only if we stop treating Jenkins logs as the whole truth.
References
[1] V. H. A. Nguyen, “An agile approach for managing microservices‑based software development: Case study in FinTech,” in Information Systems, M. Themistocleous and M. Papadaki, Eds. Cham, Switzerland: Springer, 2022, pp. 51–65. doi: 10.1007/978-3-030-95947-0_51.
[2] “Monolithic to microservices architecture A framework for design and implementation,” in Proc. IEEE Conf. (Date of Conference: 14–16 Dec. 2022), Chennai, India, 2022, pp. 1–6. (Date Added to IEEE Xplore: 23 Mar. 2023).
[3] S. Lindevall, “Agile methods in the financial services industry: A systematic literature review,” bachelor’s thesis, Arcada University of Applied Sciences, Helsinki, Finland, 2023. [Online]. Available: https://www.theseus.fi/handle/10024/800416.
[4] D. de Borba, M. S. Chaves, and M. Oliveira, “Knowledge sharing in the banking sector: A systematic literature review and research agenda,” Int. J. Knowl. Manag. Stud., vol. 13, no. 1, pp. 55–70, 2022. doi: 10.1504/IJKMS.2022.119260.
[5] V. Chang, P. Baudier, H. Zhang, Q. Xu, J. Zhang, and M. Arami, “How Blockchain can impact financial services – The overview, challenges and recommendations from expert interviewees,” Technol. Forecast. Soc. Change, vol. 158, art. no. 120166, Sep. 2020. doi: 10.1016/j.techfore.2020.120166.
[6] M. P. Robillard, “Turnover-induced knowledge loss in practice,” in *Proc. 29th ACM Joint Meeting on Eur. Softw. Eng. Conf. Symp. Found. Softw. Eng. (ESEC/FSE ’21)*, Athens, Greece, Aug. 2021, pp. 1292–1302. doi: 10.1145/3468264.3473128.
[7] “Using architecture decision records in open source projects An MSR study on GitHub,” in *Proc. IEEE/ACM 20th Int. Conf. Mining Softw. Repositories (MSR 2023)*, Melbourne, Australia, May 2023, pp. 1–6. (Date Added to IEEE Xplore: 19 Jun. 2023).
[8] “Internet financial fraud detection based on a distributed big data approach with Node2vec,” IEEE Access, vol. 9, pp. 43378–43386, 2021. doi: 10.1109/ACCESS.2021.3062467.
[9] “Aligning XAI explanations with software developers’ expectations: A case study with code smell prioritization,” Expert Syst. Appl., vol. 237, art. no. 121456, 2023. doi: 10.1016/j.eswa.2023.121456.
[10] C. Qian et al., “ChatDev: Communicative agents for software development,” in Proc. 61st Annu. Meeting Assoc. Comput. Linguistics (ACL), Toronto, ON, Canada, Jul. 2023, pp. 1–12. doi: 10.48550/arXiv.2307.07924.
[11] W. Wang, Y. Wang, S. Joty, and S. C. H. Hoi, “RAP‑Gen: Retrieval‑augmented patch generation with CodeT5 for automatic program repair,” in *Proc. 31st ACM Joint Eur. Softw. Eng. Conf. Symp. Found. Softw. Eng. (ESEC/FSE 2023)*, San Francisco, CA, USA, Dec. 2023, pp. 146–158. doi: 10.1145/3611643.3616256.
[12] A. Fan, B. Gokkaya, M. Harman, M. Lyubarskiy, S. Sengupta, S. Yoo, and J. M. Zhang, “Large language models for software engineering: Survey and open problems,” in *Proc. IEEE/ACM 45th Int. Conf. Softw. Eng.: Future of Softw. Eng. (ICSE‑FoSE)*, Melbourne, Australia, May 2023, pp. 1–15. doi: 10.1109/ICSE‑FoSE.2023.00005. (Also arXiv preprint arXiv:2310.03533, Oct. 2023).
[13] M. Noor and Z. Rana, “Natural language processing (NLP) based extraction of tacit knowledge from written communication during software development,” in Proc. 4th Int. Conf. Advancements Comput. Sci. (ICACS), Lahore, Pakistan, Feb. 2023, pp. 1–5. doi: 10.1109/ICACS55311.2023.10087762.
