A Phased Workflow for Operating LLM-Based Coding Agents
2026-08-31 • Software Engineering
Software Engineering
AI summaryⓘ
The authors studied how people work with coding agents powered by large language models (LLMs) to create software. They designed a four-step process that starts with a lot of human involvement and gradually lets the agents take over as the work progresses. A key challenge is managing context properly throughout these steps to avoid mistakes and fragile code. They found that early errors can cause bigger problems later, so careful review at the start is important. The authors also point out that there are still no good ways to measure how well these workflows work and that more practical guidance for managing context is needed.
LLM-based coding agentsworkflowcontext managementagent-assisted developmenthuman reviewerror propagationfoundation modelsoftware developmenttask delegationfailure modes
Authors
Ante Kapetanovic, Tomislav Duricic, Andro Mercep, Emanuel Lacic
Abstract
LLM-based coding agents combine a foundation model with a harness that shapes agent behavior. For non-trivial tasks, how practitioners structure their work with the coding agents determines whether reliable results follow. We report on a phased workflow for operating coding agents developed by the AI research team at Infobip. The workflow structures agent-assisted development into four phases where human effort is front-loaded and delegation increases as artifacts mature. Context management is the central concern, addressed through four strategies applied at each phase to counter known failure modes. From practitioner experience, we observe that upstream errors in research and planning can compound across later phases, while correcting generated code can introduce bloat and fragility. This motivates front-loading human review. We identify two open problems: the absence of metrics for workflow effectiveness and the gap between formalized context management components and the workflow-level patterns that practitioners need.