ACEM (Agentic Cost Estimation Model), which decomposes total agentic development cost into three additive dimensions: LLM, HITL, and infrastructure cost, is presented as a fully specified model structure and calibration methodology, with constants left symbolic pending empirical grounding.
Abstract
Traditional software cost estimation models, such as COCOMO II, Function Points, and Story Points, assume that development effort is primarily driven by human labor in design, coding, and testing. Agentic software engineering, where autonomous AI agents perform substantial implementation work and humans focus on planning, specification, and validation, challenges this assumption. New cost dimensions arise: large language model (LLM) token consumption across agent actions, Human-in-the-Loop (HITL) oversight effort, and infrastructure costs for agent orchestration and tooling. These costs are nondeterministic: identical tasks may consume different tokens, follow divergent reasoning paths, and require varying human correction, phenomena absent in traditional development. A new framework is needed to bridge standard sizing metrics with this cost structure. This paper proposes ACEM (Agentic Cost Estimation Model), which decomposes total agentic development cost into three additive dimensions: LLM, HITL, and infrastructure cost. ACEM introduces three constructs for agentic dynamics: the Revision Factor (RF), modeling token overhead from output rejection and retries; the Context Factor (CF), capturing rising token consumption as context accumulates; and the HITL Intensity Score (HIS), a four-level oversight classification scheme. It further maps Use Case Points, Story Points, and Function Points to estimated token consumption, enabling organizations to reuse existing project-scoping data for agentic cost forecasting. ACEM is presented as a fully specified model structure and calibration methodology, with constants left symbolic pending empirical grounding. As an early-stage proposal, it invites the research community to calibrate, test, and extend the model through real project data.
Developing a GenAI agent that autonomously formulates, prioritizes, and completes a limited set of code improvement tasks enterprise-wide is reported, and a Difference-in-Differences analysis found that automated improvements had an effect on the reduction of future maintenance effort that was as strong as or stronger than manual improvements confirming the feasibility of autonomous maintenance.
MAGE explains how externalized knowledge, bounded action, independent evaluation, and retained human authority can compose into a governed engineering environment, and proposes tests of when that environment turns commodity intelligence into durable engineering progress.
James C. Davis, Kelechi G. Kalu, Huiyun Peng et al.· 1 citation
This note argues that a skill is a software artefact and that its construction should follow software-engineering principles, with qualifications: single responsibility, separation of interface from implementation, low coupling, and economy in a shared token budget, together with behavioural evaluation in place of deterministic testing.
Software effort estimation remains a cornerstone of project planning and control, yet existing estimation models are grounded in the assumption that software development effort is dominated by human reasoning and manual construction. The rapid integration of large language models (LLMs) into development workflows fundamentally challenges this assumption by automating substantial portions of code synthesis while shifting human effort toward supervision, validation, and integration. As a result, traditional effort estimation proxies such as Story Points and size-based metrics may no longer reliably characterize development effort. This paper presents an empirical study examining how effort manifests in LLM-assisted software development. Rather than using LLMs as predictive estimation tools, we investigate how their adoption reshapes the underlying cost structure of development work. We introduce the notion of Hybrid Intelligence Effort (HIE), conceptualizing effort as the combined burden of model-performed reasoning and human oversight activities. Using a controlled experiment involving 22 developers, 110 real-world tasks, and three LLMs, we compare the explanatory power of traditional estimation metrics against interaction- and oversight-based Hybrid Intelligence dimensions. Our results show that while Story Points retain partial explanatory validity, they fail to capture dominant sources of effort in LLM-assisted workflows. In controlled experiments, HIE dimensions increase explained variance in observed effort from approximately 72–80%, while substantially reducing systematic estimation error. Human validation and corrective intervention emerge as the primary drivers of effort, outweighing artifact-level characteristics. These findings suggest that effort estimation models must move beyond human-centric and size-based assumptions to remain effective in AI-augmented software engineering.
Feisal Alaswad, E. Poovammal, Kadiyala Ramana et al.· Discover Computing· 1 citation· ⚡1
Agent Gym is introduced, a modular, domain-agnostic framework that wraps any existing LLM-based agent in a continuous evaluation-and-evolution loop and introduces the Spec-to-Note Gap, an autoencoder-inspired view of agentic system transparency.
Pouya Ghiasnezhad Omran, Michael Zimmermann, Duncan Cambridge et al.· 0 citations
A ScholForge - hierarchical multi-agent LLM system that tackles end-to-end software engineering research end-to-end, with a natural-language research goal, and produces working code, experimental results, and a compiled manuscript without human intervention.
A. Stepin, Boris Tolstokulakov, V. Kulikov et al.· SIGSOFT FSE Companion· 0 citations