Agentic engineering is fundamentally reshaping software development workflows, forcing practitioners to navigate the tension between frontier model capabilities and the need for granular control. Developers are increasingly skeptical of the opaque communication strategies employed by major labs like OpenAI and Anthropic, favoring transparent, open-weight models that allow for enterprise-specific fine-tuning. Effective agentic workflows now rely on minimalist harnesses and local tools—such as terminal-based multiplexers and annotation-driven code review platforms—that prevent agents from obscuring the underlying code structure. This shift toward type-driven development, where human-defined abstractions guide agentic output, mitigates the risks of non-deterministic behavior. Amidst this rapid technological evolution, the industry faces significant burnout, prompting a renewed interest in local-first technologies and sustainable, human-centric engineering practices that prioritize data sovereignty and developer autonomy over black-box automation.
Sign in to continue reading, translating and more.
Open full episode in Podwise