Specialized research software has historically been costly and time-consuming to create, but large language models (LLMs) have become capable enough at code generation to fundamentally change this. We describe how LLM-assisted programming disrupts the landscape by allowing researchers to build tools without support from software engineers, illustrate this with an example built rapidly by a single LLM-assisted developer, and discuss opportunities and risks.
This is a preview of subscription content, access via your institution
Access options
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$32.99 / 30 days
cancel any time
Subscribe to this journal
Receive 12 print issues and online access
$259.00 per year
only $21.58 per issue
Buy this article
- Purchase on SpringerLink
- Instant access to the full article PDF.
USD 39.95
Prices may be subject to local taxes which are calculated during checkout
Subjects
Code availability
MOSS is openly available under an MIT license at https://github.com/StructuralNeurobiologyLab/MOSS
References
Hocquet, A. et al. Nat. Comput. Sci. 4, 465–468 (2024).
Schindelin, J. et al. Nat. Methods 9, 676–682 (2012).
Berg, S. et al. Nat. Methods 16, 1226–1232 (2019).
Sofroniew, N. et al. napari. Zenodo https://doi.org/10.5281/ZENODO.3555620 (2019).
Makovetsky, R., Piché, N. & Marsh, M. Microsc. Microanal. 24, 532–533 (2018). (Suppl. 1).
Last, M. G. F., Abendstein, L., Voortman, L. M. & Sharp, T. H. eLife 13, RP98552 (2024).
Ouyang, W. et al. Preprint at bioRxiv https://doi.org/10.1101/2022.06.07.495102 (2022).
Archit, A. et al. Nat. Methods 22, 579–591 (2025).
OpenAI et al. Preprint at arXiv https://doi.org/10.48550/arXiv.2303.08774 (2023).
Jimenez, C. E. et al. in International Conference on Learning Representations 54107–54157 (2024).
Peng, S. et al. Preprint at arXiv https://doi.org/10.48550/arXiv.2302.06590 (2023).
Karpathy, A. X https://x.com/karpathy/status/1886192184808149383 (2025).
Ronneberger, O., Fischer, P. & Brox, T. in Medical Image Computing and Computer-Assisted Intervention 234–241 (Springer, 2015).
Jumper, J. et al. Nature 596, 583–589 (2021).
Kirillov, A. et al. in Proc. IEEE/CVF International Conference on Computer Vision 3992–4003 (IEEE, 2023).
Acknowledgements
MOSS was developed using LLM-assisted programming tools including Anthropic’s Claude. We thank M. So-Last for helpful discussions on U-Nets.
Author information
Authors and Affiliations
Contributions
MOSS was developed by N.D.M. The manuscript was written jointly by N.D.M and J.M.R.K.
Corresponding author
Ethics declarations
Competing interests
J.M.R.K. owns shares of ariadne.ai ag.
Peer review
Peer review information
Nature Methods thanks Robert Haase, Wei Ouyang and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Rights and permissions
About this article
Cite this article
Medina, N.D., Kornfeld, J.M.R. Disruption of the research software landscape through AI software generation. Nat Methods (2026). https://doi.org/10.1038/s41592-026-03210-x
Published:
Version of record:
DOI: https://doi.org/10.1038/s41592-026-03210-x
Facts Only
* Specialized research software was historically costly and time-consuming to create.
* Large language models (LLMs) are capable of code generation.
* LLM-assisted programming allows researchers to build tools without support from software engineers.
* A rapid example of tool building was demonstrated by a single LLM-assisted developer.
* MOSS is openly available under an MIT license at https://github.com/StructuralNeurobiologyLab/MOSS
* MOSS was developed using LLM-assisted programming tools including Anthropic’s Claude.
Executive Summary
Full Take
Sentinel — Human
The text reads like the introductory framing for a research article, characterized by direct argumentation and established academic citation practices.