Intelligent CI/CD at Scale: Autonomous Agents Transforming Software Delivery Pipelines
Abstract
Most CI/CD pipelines treat every change the same way. A one-line config fix runs the same full test suite as a core refactor, and a change that breaks a consumer on an event bus can pass every test, because no static dependency graph shows that consumer.
I built a framework that gives the pipeline a way to judge risk. Four agents split the work. One works out what a commit changed and whether it breaks an API contract. One finds the services that depend on it, including those reached through queues and shared databases. One runs only the tests those services need. The last picks how to release: low-risk changes ship on their own, medium-risk changes go out as a canary with automatic rollback, and high-risk changes go blue-green with a person approving the switch. Every decision is logged, anyone can override it, and new behavior runs in shadow mode first. It sits on top of existing pipeline tools.
I applied it to the 21 services my teams own. We set up the first two by hand, then automated onboarding, and the other 19 came on without manual setup. All 21 now run on fully automated CI/CD.
- Published in
- Proceedings of the Raptors Conference, Vol. 1 (2026)
- Conference
- Raptors Conference 2026, 21 October 2026, online
- Accepted
- 2026-09-16
- DOI
- https://doi.org/10.68495/rc.2026.C032
- Presentation
- Slides (PDF)
- Licence
- © 2026Karishma Verma. Published under Creative Commons Attribution 4.0 International (CC BY 4.0).
- Publisher
- Hackathon Raptors, Community Interest Company 15557917, London, United Kingdom
- Permanent address
- https://proceedings.raptors.dev/verma-intelligent-ci-cd-at
Open access. Every entry in the Proceedings is free to read, with no charge to authors and no charge to readers. Slides are served from this site, so they stay available at the paper's permanent address.
Cite this paper
Karishma Verma (2026). Intelligent CI/CD at Scale: Autonomous Agents Transforming Software Delivery Pipelines. Proceedings of the Raptors Conference, Vol. 1 (2026). Hackathon Raptors, London. https://doi.org/10.68495/rc.2026.C032
@inproceedings{raptors2026-c032,
title = {{Intelligent CI/CD at Scale: Autonomous Agents Transforming Software Delivery Pipelines}},
author = {Verma, Karishma},
booktitle = {Proceedings of the Raptors Conference},
volume = {1},
year = {2026},
publisher = {Hackathon Raptors},
address = {London, United Kingdom},
doi = {10.68495/rc.2026.C032},
url = {https://doi.org/10.68495/rc.2026.C032},
abstract = {Most CI/CD pipelines treat every change the same way. A one-line config fix runs the same full test suite as a core refactor, and a change that breaks a consumer on an event bus can pass every test, because no static dependency graph shows that consumer. I built a framework that gives the pipeline a way to judge risk. Four agents split the work. One works out what a commit changed and whether it breaks an API contract. One finds the services that depend on it, including those reached through queues and shared databases. One runs only the tests those services need. The last picks how to release: low-risk changes ship on their own, medium-risk changes go out as a canary with automatic rollback, and high-risk changes go blue-green with a person approving the switch. Every decision is logged, anyone can override it, and new behavior runs in shadow mode first. It sits on top of existing pipeline tools. I applied it to the 21 services my teams own. We set up the first two by hand, then automated onboarding, and the other 19 came on without manual setup. All 21 now run on fully automated CI/CD.}
}
TY - CONF TI - Intelligent CI/CD at Scale: Autonomous Agents Transforming Software Delivery Pipelines AU - Verma, Karishma T2 - Proceedings of the Raptors Conference VL - 1 PY - 2026 DA - 2026/09/16 PB - Hackathon Raptors CY - London, United Kingdom DO - 10.68495/rc.2026.C032 UR - https://doi.org/10.68495/rc.2026.C032 AB - Most CI/CD pipelines treat every change the same way. A one-line config fix runs the same full test suite as a core refactor, and a change that breaks a consumer on an event bus can pass every test, because no static dependency graph shows that consumer. I built a framework that gives the pipeline a way to judge risk. Four agents split the work. One works out what a commit changed and whether it breaks an API contract. One finds the services that depend on it, including those reached through queues and shared databases. One runs only the tests those services need. The last picks how to release: low-risk changes ship on their own, medium-risk changes go out as a canary with automatic rollback, and high-risk changes go blue-green with a person approving the switch. Every decision is logged, anyone can override it, and new behavior runs in shadow mode first. It sits on top of existing pipeline tools. I applied it to the 21 services my teams own. We set up the first two by hand, then automated onboarding, and the other 19 came on without manual setup. All 21 now run on fully automated CI/CD. LA - en ER -