Pith. sign in

REVIEW 3 major objections 5 minor 35 references

Tether: A Personalized Support Assistant for Software Engineers with ADHD

T0 review · 3 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Tether is an LLM-powered desktop assistant that monitors developer activity, detects disengagement, and nudges software engineers with ADHD back into focus using RAG-grounded, ADHD-informed messages.

desk verdict Novel, honestly-described ADHD support tool for developers; the prototype is real, but the paper's only empirical claim—improved contextual accuracy from self-use—has no numbers behind it. read the letter →

arxiv 2509.01946 v1 pith:WRVNU4Y5 submitted 2025-09-02 cs.SE

Aarsh Shah , Cleyton Magalhaes , Kiev Gama , Ronnie de Souza Santos This is my paper · ORCID
classification cs.SE
keywords ADHDsoftwareengineeringLLMretrieval-augmentedgenerationassistivetoolsgamificationneurodiversityactivitymonitoring
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper presents Tether, a desktop application that combines an LLM chatbot, local OS activity monitoring, and gamification to help software engineers with ADHD maintain focus and re-engage after distraction. The authors are trying to establish that a tool embedded in the developer's own environment can deliver adaptive, ADHD-informed support that generic focus apps and browser blockers do not. Their preliminary validation is a self-use evaluation reporting improved contextual accuracy after iterative refinements to prompts and the retrieval pipeline, plus a feature comparison with existing ADHD tools. If the central claim holds, LLM-powered context-aware assistants become a viable direction for neurodiversity-aware tooling in software engineering, although effectiveness with actual target users remains untested.

What carries the argument

The central mechanism is the monitoring-to-prompt loop. An OS-level tracker records active windows, idle time, and recovery patterns; a modular prompt template turns those signals plus chat history and retrieved ADHD-specific material into an instruction for the LLM to produce a short re-engagement nudge or a conversational reply. Gamification closes the loop by rewarding sustained focus, efficient task switches, and quick recovery from distraction.

What would settle it

Run Tether while a developer deliberately reads a long error trace or design document without touching the keyboard; if the tool sends a re-engagement notification each time, the core engagement-detection premise fails. A diary study comparing Tether's nudge timestamps with self-reported focused periods would quantify this false-positive rate.

Watch

Extended reading notes

Core claim

Tether's design loops four components together: a conversational bot, LLM response generation, a retrieval-augmented generation (RAG) pipeline that indexes ADHD-specific literature and prior user interactions, and a gamification engine that awards points and badges for focus behavior. A local monitoring engine tracks active windows, idle time, and recovery patterns; when it detects disengagement, the backend builds a structured prompt from recent activity, chat history, and metadata, sends it to an external LLM, and routes the output either to an OS notification nudge or to the chatbot. The paper's central claim is that this loop gives real-time, personalized support grounded in ADHD strateg

Load-bearing premise

The system assumes that OS-level activity—active window and idle time—reliably indicates whether a developer is engaged; productive inactivity like reading or thinking is indistinguishable from distraction, so the trigger may misfire.

Editorial extensions

If this is right

  • If Tether works as claimed, a developer who goes idle receives a supportive, context-specific nudge instead of a generic timer, and can ask the chatbot for ADHD-informed task breakdowns without leaving the development environment.
  • The architecture—local activity signals plus a RAG pipeline plus modular prompts—can be reproduced by other teams, with all user data except the LLM calls kept on the machine.
  • The feature comparison implies existing ADHD apps (browser blockers, cognitive trainers, wearables) do not combine monitoring, chat, and development-awareness, leaving a niche Tether occupies.
  • The planned studies with healthcare professionals and with software engineers with ADHD would be the next direct test of whether the approach actually changes focus behavior in daily workflows.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The idle-time trigger is the fragile link: until gaze or audio sensing is added, Tether cannot distinguish deep work from disengagement, and mistimed nudges could add friction; the authors acknowledge this gap and plan additional sensing channels.
  • A concrete extension would be a diary study that logs Tether's nudge timestamps against self-reported 'in focus' periods, quantifying how often the OS-level monitor fires during productive inactivity.
  • If the pattern proves effective for ADHD, the same local-monitoring-plus-RAG design could generalize to other executive-function conditions or to any knowledge worker whose attention benefits from external scaffolding.
  • The gamification layer itself invites a behavioral question the paper does not answer: whether points, badges, and unlockable themes help or hinder intrinsic motivation for ADHD users over time.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper introduces Tether, an LLM-powered desktop application intended to support software engineers with ADHD through context-aware nudges, a RAG-grounded chatbot, and gamification. The authors argue that existing ADHD support tools are not tailored to high-focus technical work, and they propose a prototype that combines local OS activity monitoring, a LangChain RAG pipeline over ADHD literature, and LLM-generated notifications and chat responses. The stated preliminary validation is a self-use evaluation that reportedly showed improved contextual accuracy after prompt and RAG refinements, plus a feature-level comparison against existing tools. The paper also includes a discussion of threats to validity, future plans for gaze/microphone sensing, and two planned user studies.

Significance. If the empirical claims were supported, Tether would address a genuinely underserved intersection: neurodivergent software engineers and LLM-based assistive tools. The paper's strengths are its identification of a real gap, its grounding in clinical principles for externalizing executive functions, and its public release of a working prototype with architectural details. However, the evidence presented is anecdotal and self-referential. The central validation claim is not accompanied by any measurements, evaluation protocol, or independent assessment, so the current significance is limited to that of a design-and-prototype report. The contribution is plausible and worth pursuing, but the paper does not yet substantiate the headline assertion of improved contextual accuracy.

major comments (3)
  1. [Abstract and Section III-B/IV-C] The abstract's claim that 'Preliminary validation through self-use revealed improved contextual accuracy following iterative prompt refinements and RAG enhancements' is not supported by the paper. Section III-B explicitly states 'No formal user study was conducted during this stage,' and Section IV-C offers only a feature-level comparison (Table I). There is no definition of 'contextual accuracy,' no before/after measurements, no sample prompts or responses, no rubric, and no independent evaluator. Because this claim is the paper's only reported evaluation outcome, it is load-bearing. The authors should either remove or substantially qualify this claim, or provide the missing evidence: e.g., a small protocol with annotated examples, scoring criteria, number of iterations, and raw logs.
  2. [Section IV-C / Table I] The comparison in Table I does not constitute validation of Tether's effectiveness. The dimensions (Monitoring, Chat, Dev-Aware, RAG, Gamified) are chosen by the authors, and the Yes/No entries for third-party tools are self-assessed with no sources or operational definitions. For instance, 'Monitoring' is undefined (does it imply passive behavioral sensing?), and 'Dev-Aware' is not clearly distinct from general task support. If this table is presented as the preliminary evaluation, the authors must clarify that it is a design-space comparison, not an empirical result. At minimum, each column should be defined and each entry should be justified or cited.
  3. [Section III-A and Section IV-B] The disengagement detection mechanism equates OS-level idle time and active-window loss with loss of focus. A developer may be productively inactive while reading a stack trace, thinking through a design, or using a second monitor. The paper acknowledges this in Section VII ('to better distinguish between focused inactivity and disengagement'), but the notification trigger is a core part of Tether's value proposition. Without a sensitivity analysis, a discussion of idle thresholds, or a small feasibility test, the design claim that Tether provides adaptive context-aware support is weakened. The authors should address this proxy concretely, even if only by stating the default thresholds and their rationale.
minor comments (5)
  1. [Section I] The reference [10] is duplicated in two places ('[3], [10], [10], [12]' and '[3], [10], [10]–[12]'). Please clean up duplicate citations.
  2. [Section IV-C] The text says the comparison explored 'engagement, interaction, and gamification,' but Table I has no 'engagement' or 'interaction' columns. Either align the text with the table or add the missing columns.
  3. [Section V] The phrase 'the systematic literature reviews cited in Section 2' should be singular ('systematic literature review') unless two are meant; the cited sources do not all appear to be systematic reviews.
  4. [References] Reference [34] is missing a year and venue. Please complete the bibliographic entry.
  5. [Section III-B] The engineering research methodology is mentioned in the abstract and Section III, but the specific 'evaluation' phase is limited to a comparison with literature. Consider labelling Section III-B as 'Design-space comparison' rather than 'Preliminary evaluation' to avoid overstatement.

Circularity Check

1 steps flagged · score 4.0 of 10

Central 'improved contextual accuracy' claim is a self-referential tuning loop; the design itself is not circular.

  1. fitted input called prediction [Abstract; Section III-B; Section VI]
    "Preliminary validation through self-use revealed improved contextual accuracy following iterative prompt refinements and RAG enhancements. ... No formal user study was conducted during this stage. ... this preliminary version of our tool was tested in a simulated setting with predefined scenarios, while it was being developed, which may not fully capture the needs and behaviors of professionals with ADHD in real-world development environments."

    The abstract's only empirical result is that after the authors iteratively refined prompts and RAG enhancements, self-use showed improved 'contextual accuracy.' The refinements are the tuned parameters and the self-use sessions are the evaluation data; no metric, baseline, rubric, external evaluator, or formal user study is reported. The people who made the changes are the same people who judged the improvement, so the claimed improvement is a restatement of the tuning loop ('we changed the system and, by our own assessment, it got better') rather than an independent prediction. This fits the fitted-input-called-prediction pattern: the evaluation target is the same quantity the tuning was aimed at, so the headline empirical claim reduces to self-assessment.

full rationale

Tether is a design report with no formal derivation chain; there are no equations whose output equals an input. The self-citations ([1], [10], [11]) are used only for background on ADHD in software engineering and do not carry the validity of the tool. The acknowledged limitations (no user study, simulated testing, no expert validation) are stated candidly and mitigate overreach. The one circular element is the validation loop: the central claim of improved contextual accuracy is produced by the same iterative refinement process it is supposed to validate, with no separation between tuning and measurement. That is a partial circularity in the evaluation, not in the artifact design. Score 4 reflects one self-referential validation claim while the core architectural contribution remains independent.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No numeric free parameters are fit in this design paper; the load-bearing assumptions are qualitative domain assumptions about ADHD support, LLM benefits, gamification, and the validity of OS-level monitoring as a focus proxy. The paper relies on prior literature for the first three and on an untested heuristic for the fourth.

assumptions (4)
  • domain assumption Externalizing executive functions through environmental scaffolding improves task performance for adults with ADHD.
    Cited from Fleming and McMahon [25] and used throughout as the design principle motivating Tether's monitoring and prompts (Section III).
  • domain assumption LLM-based conversational support reduces cognitive overhead and increases engagement for ADHD users.
    Borrowed from references [30]-[33] and used to justify the chatbot interface, but not tested for this population in this paper.
  • domain assumption Gamification with points, badges, and unlockable themes reinforces sustained engagement for ADHD adults.
    Supported by references [21], [22], [24], but with no direct evidence in the target population of software engineers; the paper adopts it as a motivating feature.
  • domain assumption Local activity monitoring signals (active window, idle time) are a valid proxy for focus and disengagement.
    Adopted in Section III-A without validation; the paper acknowledges in Future Work that focused inactivity (reading, thinking) is not distinguished, so this assumption is load-bearing and currently unverified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Tether: A Personalized Support Assistant for Software Engineers with ADHD." pith.science (2026). https://pith.science/paper/WRVNU4Y5

@misc{pith2026250901946,
  author       = {Pith},
  title        = {Pith review of: Tether: A Personalized Support Assistant for Software Engineers with ADHD},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WRVNU4Y5}},
  note         = {Machine review of arXiv:2509.01946}
}
read the original abstract

Equity, diversity, and inclusion in software engineering often overlook neurodiversity, particularly the experiences of developers with Attention Deficit Hyperactivity Disorder (ADHD). Despite the growing awareness about that population in SE, few tools are designed to support their cognitive challenges (e.g., sustained attention, task initiation, self-regulation) within development workflows. We present Tether, an LLM-powered desktop application designed to support software engineers with ADHD by delivering adaptive, context-aware assistance. Drawing from engineering research methodology, Tether combines local activity monitoring, retrieval-augmented generation (RAG), and gamification to offer real-time focus support and personalized dialogue. The system integrates operating system level system tracking to prompt engagement and its chatbot leverages ADHD-specific resources to offer relevant responses. Preliminary validation through self-use revealed improved contextual accuracy following iterative prompt refinements and RAG enhancements. Tether differentiates itself from generic tools by being adaptable and aligned with software-specific workflows and ADHD-related challenges. While not yet evaluated by target users, this work lays the foundation for future neurodiversity-aware tools in SE and highlights the potential of LLMs as personalized support systems for underrepresented cognitive needs.

Figures

Figures reproduced from arXiv: 2509.01946 by the authors.

Figure 3
Figure 3. Main user interface and chatbot elements. [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 1
Figure 1. System architecture of the Tether application. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 4
Figure 4. Gamification aspects. B. Internal Process Tether’s backend processes are initiated either through pas￾sive detection (e.g., user inactivity) or active engagement 3 [PITH_FULL_IMAGE:figures/full_fig_p003_4.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 32 canonical work pages

  1. [1]

    Explor- ing hybrid work realities: A case study with software professionals from underrepresented groups,

    R. d. S. Santos, C. Magalhaes, R. Santos, and J. Correia-Neto, “Explor- ing hybrid work realities: A case study with software professionals from underrepresented groups,” in Companion Proceedings of the 32nd ACM International Conference on the Foundations of Software Engineering , 2024, pp. 27–37

    work page 2024

  2. [2]

    Inclusion of individuals with autism spectrum disorder in software engineering,

    G. M ´arquez, M. Pacheco, H. Astudillo, C. Taramasco, and E. Calvo, “Inclusion of individuals with autism spectrum disorder in software engineering,” Information and Software Technology , p. 107434, 2024

    work page 2024

  3. [3]

    “i felt pressured to give 100% all the time

    N. da Silva Menezes, T. ´A. da Rocha, L. S. S. Camelo, and M. P. Mota, ““i felt pressured to give 100% all the time”: How are neurodi- vergent professionals being included in software development teams?” in Simp´osio Brasileiro de Sistemas de Informac ¸˜ao (SBSI). SBC, 2025, pp. 525–534

    work page 2025

  4. [4]

    Differences between Neurodivergent and Neurotypical Software Engineers: Analyzing the 2022 Stack Overflow Survey

    P. Verma, M. V . Cruz, and G. Liebel, “Differences between neurodi- vergent and neurotypical software engineers: Analyzing the 2022 stack overflow survey,” arXiv preprint arXiv:2506.03840 , 2025

    work page Pith review arXiv 2022

  5. [5]

    Perceived diversity in software engineering: a systematic literature review,

    G. Rodr ´ıguez-P´erez, R. Nadri, and M. Nagappan, “Perceived diversity in software engineering: a systematic literature review,”Empirical Software Engineering, vol. 26, pp. 1–38, 2021

    work page 2021

  6. [6]

    The diversity crisis in software development,

    K. Albusays, P. Bjorn, L. Dabbish, D. Ford, E. Murphy-Hill, A. Sere- brenik, and M.-A. Storey, “The diversity crisis in software development,” IEEE Software, vol. 38, no. 2, pp. 19–25, 2021

    work page 2021

  7. [7]

    A systematic mapping study of diversity in software engineering: a perspective from the agile methodologies,

    K. K. Silveira and R. Prikladnicki, “A systematic mapping study of diversity in software engineering: a perspective from the agile methodologies,” in 2019 IEEE/ACM 12th International Workshop on Cooperative and Human Aspects of Software Engineering (CHASE) . IEEE, 2019, pp. 7–10

    work page 2019

  8. [8]

    “get me in the groove

    K. Newman, S. Snay, M. Endres, M. Parikh, and A. Begel, ““get me in the groove”: A mixed methods study on supporting ADHD professional programmers,” in 2025 IEEE/ACM 47th Intl Conference on Software Engineering (ICSE). IEEE Computer Society, 2025, pp. 778–778

    work page 2025

Show all 35 references
  1. [9]

    Stack overflow developer survey 2022,

    Stack Overflow, “Stack overflow developer survey 2022,” 2022, accessed: 2025-06-24. [Online]. Available: https://survey.stackoverflow. co/2022/

    2022

  2. [10]

    A socio- technical grounded theory on the effect of cognitive dysfunctions in the performance of software developers with ADHD and autism,

    K. Gama, G. Liebel, M. Goul ˜ao, A. Lacerda, and C. Lacerda, “A socio- technical grounded theory on the effect of cognitive dysfunctions in the performance of software developers with ADHD and autism,” in 2025 IEEE/ACM 47th International Conference on Software Engineering: Sof...

    2025

  3. [11]

    Challenges, strengths, and strategies of software engineers with adhd: A case study,

    G. Liebel, N. Langlois, and K. Gama, “Challenges, strengths, and strategies of software engineers with adhd: A case study,” inProceedings of the 46th International Conference on Software Engineering: Software Engineering in Society , 2024, pp. 57–68

    2024

  4. [12]

    Understanding the chal- lenges faced by neurodiverse software engineering employees: Towards a more inclusive and productive technical workforce,

    M. R. Morris, A. Begel, and B. Wiedermann, “Understanding the chal- lenges faced by neurodiverse software engineering employees: Towards a more inclusive and productive technical workforce,” in Proceedings of the 17th International ACM SIGACCESS Conference on computers & acces...

    2015

  5. [13]

    Future of ADHD care: Evaluating the efficacy of chatgpt in therapy enhancement,

    S. Berrezueta-Guzman, M. Kandil, M.-L. Mart ´ın-Ruiz, I. Pau de la Cruz, and S. Krusche, “Future of ADHD care: Evaluating the efficacy of chatgpt in therapy enhancement,” in Healthcare, vol. 12, no. 6. MDPI, 2024, p. 683

    2024

  6. [14]

    Integrating ai into ADHD therapy: Insights from chatgpt-4o and robotic assistants,

    S. Berrezueta-Guzman, M. Kandil, and S. Wagner, “Integrating ai into ADHD therapy: Insights from chatgpt-4o and robotic assistants,” Human-Centric Intelligent Systems , pp. 1–16, 2025

    2025

  7. [15]

    Applying large language models to assess quality of care: Monitoring ADHD medication side effects,

    Y . Bannett, F. Gunturkun, M. Pillai, J. E. Herrmann, I. Luo, L. C. Huffman, and H. M. Feldman, “Applying large language models to assess quality of care: Monitoring ADHD medication side effects,” Pediatrics, vol. 155, no. 1, p. e2024067223, 2025

    2025

  8. [16]

    ADHD in ehealth-a systematic literature review,

    M. R. Olsen, C. Casado-Lumbreras, and R. Colomo-Palacios, “ADHD in ehealth-a systematic literature review,” Procedia Computer Science , vol. 100, pp. 207–214, 2016

    2016

  9. [17]

    Design and evaluation of llm-based conversa- tional virtual characters to assist adults with ADHD,

    T. Li, X. Hu, and X. Xu, “Design and evaluation of llm-based conversa- tional virtual characters to assist adults with ADHD,” in Intl Conference on Human-Computer Interaction . Springer, 2025, pp. 360–376

    2025

  10. [18]

    Exploring large language models through a neurodivergent lens: Use, challenges, community- driven workarounds, and concerns,

    B. Carik, K. Ping, X. Ding, and E. H. Rho, “Exploring large language models through a neurodivergent lens: Use, challenges, community- driven workarounds, and concerns,” Proceedings of the ACM on Human- Computer Interaction, vol. 9, no. 1, pp. 1–28, 2025

    2025

  11. [19]

    Artificial intelligence in ADHD: a global perspective on research hotspots, trends and clinical applications,

    X. Wang, Q. Jia, L. Liang, W. Zhou, W. Yang, and J. Mu, “Artificial intelligence in ADHD: a global perspective on research hotspots, trends and clinical applications,” Frontiers in Human Neuroscience, vol. 19, p. 1577585, 2025

    2025

  12. [20]

    Mobile applications for students with ADHD,

    E. Kyriakaki and A. M. Driga, “Mobile applications for students with ADHD,” Global Journal of Engineering and Technology Advances , vol. 15, no. 3, pp. 205–216, 2023

    2023

  13. [21]

    Managing ADHD symptoms in children through the use of various technology-driven serious games: A systematic review,

    A. Doulou, P. Pergantis, A. Drigas, and C. Skianis, “Managing ADHD symptoms in children through the use of various technology-driven serious games: A systematic review,” Multimodal Technologies and Interaction, vol. 9, no. 1, p. 8, 2025

    2025

  14. [22]

    Attention- deficit/hyperactivity disorder mobile apps: A systematic review,

    C. R. P ˘as˘arelu, G. Andersson, and A. Dobrean, “Attention- deficit/hyperactivity disorder mobile apps: A systematic review,” Inter- national journal of medical informatics , vol. 138, p. 104133, 2020

    2020

  15. [23]

    ADHD: is there an app for that? a suitability assessment of apps for the parents of children and young people with ADHD,

    L. Powell, J. Parker, V . Harpin et al., “ADHD: is there an app for that? a suitability assessment of apps for the parents of children and young people with ADHD,” JMIR mHealth and uHealth , vol. 5, no. 10, p. e7941, 2017

    2017

  16. [24]

    A decade of apps for ADHD management: a scoping review,

    J. Hernandez-Capistran, G. Alor-Hernandez, L. N. Sanchez-Morales, and I. Machorro-Cano, “A decade of apps for ADHD management: a scoping review,” Behaviour & Information Technology , pp. 1–28, 2025

    2025

  17. [25]

    Developmental context and treat- ment principles for adhd among college students,

    A. P. Fleming and R. J. McMahon, “Developmental context and treat- ment principles for adhd among college students,” Clinical child and family psychology review, vol. 15, pp. 303–329, 2012

    2012

  18. [26]

    Mobile applications as inter- vention tools for children with ADHD for a sustainable education

    A. Doulou, A. Drigas, and C. Skianis, “Mobile applications as inter- vention tools for children with ADHD for a sustainable education.” Technium Sustainability, vol. 2, no. 4, pp. 44–62, 2022

    2022

  19. [27]

    Say-it & learn: Interactive application for children with ADHD,

    S. Butt, F. E. Hannan, M. Rafiq, I. Hussain, C. N. Faisal, and W. Younas, “Say-it & learn: Interactive application for children with ADHD,” in Cross-Cultural Design. Applications in Health, Learning, Communica- tion, and Creativity: 12th Intl Conference, CCD 2020, Held as Part...

    2020

  20. [28]

    Evaluation of the effectiveness of the focus ADHD app in monitoring adults with attention-deficit/hyperactivity disorder,

    L. R. Carvalho, L. M. Haas, G. Zeni, M. M. Victor, S. P. Techele, J. M. Castanho, I. M. Coimbra, A. d. F. de Sousa, N. Ceretta, A. Garrudoet al., “Evaluation of the effectiveness of the focus ADHD app in monitoring adults with attention-deficit/hyperactivity disorder,” Europea...

    2023

  21. [29]

    Empirical standards for software engineering research,

    P. Ralph, N. bin Ali, S. Baltes, D. Bianculli, J. Diaz, Y . Dittrich, N. Ernst, M. Felderer, R. Feldt, A. Filieri, B. B. N. de Franc ¸a, C. A. Furia, G. Gay, N. Gold, D. Graziotin, P. He, R. Hoda, N. Juristo, B. Kitchenham, V . Lenarduzzi, J. Mart ´ınez, J. Melegati, D. Mendez...

    2020

  22. [30]

    Bots in software engineering: a systematic mapping study,

    S. Santhanam, T. Hecking, A. Schreiber, and S. Wagner, “Bots in software engineering: a systematic mapping study,” PeerJ Computer Science, vol. 8, p. e866, Feb. 2022. [Online]. Available: https: //doi.org/10.7717/peerj-cs.866

    2022 doi

  23. [31]

    Software bots in software engineering: benefits and challenges,

    M. Wessel, M. A. Gerosa, and E. Shihab, “Software bots in software engineering: benefits and challenges,” in Proc. of the 19th International Conference on Mining Software Repositories , ser. MSR ’22. New York, NY , USA: Association for Computing Machinery, 2022, p. 724–725. [O...

    2022

  24. [32]

    Large language models for software engineering: Sur- vey and open problems,

    A. Fan, B. Gokkaya, M. Harman, M. Lyubarskiy, S. Sengupta, S. Yoo, and J. M. Zhang, “Large language models for software engineering: Sur- vey and open problems,” in 2023 IEEE/ACM International Conference on Software Engineering: Future of Software Engineering (ICSE-FoSE). IEEE...

    2023

  25. [33]

    How can recommender systems benefit from large language models: A survey,

    J. Lin, X. Dai, Y . Xi, W. Liu, B. Chen, H. Zhang, Y . Liu, C. Wu, X. Li, C. Zhu et al., “How can recommender systems benefit from large language models: A survey,” ACM Transactions on Information Systems, vol. 43, no. 2, pp. 1–47, 2025

    2025

  26. [34]

    Retrieval-augmented generation for large language models: A survey

    Y . Gao, Y . Xiong, X. Gao, K. Jia, J. Pan, Y . Bi, J. Sun, and H. Wang, “Retrieval-augmented generation for large language models: A survey.”

  27. [35]

    Understand what llm needs: Dual preference alignment for retrieval- augmented generation,

    G. Dong, Y . Zhu, C. Zhang, Z. Wang, J.-R. Wen, and Z. Dou, “Understand what llm needs: Dual preference alignment for retrieval- augmented generation,” in Proceedings of the ACM on Web Conference 2025, 2025, pp. 4206–4225. 5

    2025

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.