Edge Rewrite
// HTMLRewriter · presentation

This page was redesigned at the edge.

Cloudflare fetched the original article and streamed it through HTMLRewriter to apply an entirely new visual system without rebuilding the source page.

// request.cf · coarse context

A page that knows where it met you.

Only coarse request metadata is shown. This demo does not display or persist visitor IP addresses.

Country
US
Cloudflare location
CMH
Connection
HTTP/2
Language
Not provided

Ray ID: a41083432ba3f689

Jump to content

Draft:Agentic software engineering

From Wikipedia, the free encyclopedia

Agentic software engineering is an emerging area of software engineering concerned with the use of autonomous, goal-directed agents built on large language models to carry out software development tasks, and with the engineering of such agent systems themselves.[1][2] Whereas AI-assisted software development applies AI tools within existing development workflows, agentic software engineering has been framed as requiring a "radical reimagining" of the discipline's foundational elements — its actors, processes, tools, and artifacts — around agents pursuing goal-oriented objectives.[2]

Background

[edit]

The application of software agents to engineering problems predates large language models; agent-oriented software engineering developed around multi-agent systems and associated design methodologies from the 1990s onward.[3] Interest in the area was renewed following the introduction of large language models capable of multi-step tool use, which enabled agents to operate across tasks including requirements analysis, coding, testing, and deployment.[4][1]

Research

[edit]

A systematic literature review published in IEEE Access in 2026 surveyed methods and techniques in the area, drawing on 61 studies published between 2022 and 2025 and identifying approaches based on autonomous coding, multi-agent systems, iterative refinement, and human–agent collaboration.[1] The review applies an autonomy rating across the studies it surveys, spanning approaches from individual AI-assisted tools to fully autonomous multi-agent systems, and reports that industrial adoption has generally involved narrower agent roles rather than fully autonomous end-to-end workflows.[1] Proposals in the area have framed the shift as requiring a reorganization of software engineering's foundational elements — its actors, processes, tools, and artifacts — rather than the addition of tools within existing workflows.[2][5] Other research addresses agent architectures and coordination,[3] questions of trust and verification in agent-produced software,[6] and proposed research roadmaps for the field.[2] In April 2026, the ACM and IEEE Computer Society sponsored a workshop on the subject, co-located with the International Conference on Software Engineering in Rio de Janeiro.[7]

Terminology

[edit]

Terminology in the field is not settled, and competing variants are in concurrent use.[8] In February 2026, Andrej Karpathy proposed the term agentic engineering for work in which developers orchestrate coding agents and act as oversight rather than writing code directly, in contrast to vibe coding.[9] The 2026 ACM/IEEE workshop was titled "Agentic Engineering", while papers presented at it generally used the phrase "agentic software engineering".[7][5]

References

[edit]
  1. 1 2 3 4 Otoum, Nesreen; Elkhalili, Nuha (2026). "Methods and Techniques of Agentic Software Engineering: A Systematic Literature Review". IEEE Access. 14: 7443–7465. doi:10.1109/ACCESS.2026.3652325.
  2. 1 2 3 4 Hassan, Ahmed E.; Li, Hao; Lin, Dayi; et al. (2025). "Agentic Software Engineering: Foundational Pillars and a Research Roadmap". arXiv:2509.06216 [cs.SE].
  3. 1 2 He, Junda; Treude, Christoph; Lo, David (2025). "LLM-Based Multi-Agent Systems for Software Engineering: Literature Review, Vision, and the Road Ahead". ACM Transactions on Software Engineering and Methodology. 34 (5): 1–30. doi:10.1145/3712003.
  4. ↑ Wang, Yanlin; Zhong, Wanjun; Huang, Yanxian; et al. (2025). "Agents in software engineering: survey, landscape, and vision". Automated Software Engineering. 32 (2). doi:10.1007/s10515-025-00544-2.
  5. 1 2 Feldt, Robert; Lenberg, Per; Frattini, Julian; Parthasarathy, Dhasarathy (2026). "The Semi-Executable Stack: Agentic Software Engineering and the Expanding Scope of SE". arXiv:2604.15468 [cs.SE].
  6. ↑ Roychoudhury, Abhik; Păsăreanu, Corina; Pradel, Michael; Ray, Baishakhi (2026). "Agentic AI Software Engineers: Programming with Trust". Communications of the ACM. 69 (5): 56–58. doi:10.1145/3769314.
  7. 1 2 "AGENT '26: International Workshop on Agentic Engineering". International Conference on Software Engineering 2026. Retrieved July 20, 2026.
  8. ↑ Hoda, Rashina (2025). "Toward Agentic Software Engineering Beyond Code: Framing Vision, Values, and Vocabulary". arXiv:2510.19692 [cs.SE].
  9. ↑ Taft, Darryl K. (February 10, 2026). "Vibe Coding Is Passé. Karpathy Has a New Name for the Future of Software". The New Stack. Retrieved July 20, 2026.

Category:Software engineering Category:Artificial intelligence Category:Large language models