An eval pipeline was implemented to enhance end-to-end testing in CI/CD workflows, addressing challenges with non-determinism and the need for prompt tuning.
This post details a method for improving internal workflows by using progressive disclosure to manage large files effectively within agent systems.
Insights on facilitating AI adoption at Imprint highlight the importance of leadership, cross-team collaboration, and effective documentation in leveraging LLM-tooling and agents.
The post explores the transformation of software development through continuous deployment and coding agents, highlighting the critical roles of judgment and creativity in the process.
Effective logging is crucial for debugging internal agents, evolving from basic AWS logs to advanced Datadog integration for better accessibility and insight.
A refined approach to context window compaction in internal workflows enhances the functionality of ChatGPT 4.1 agents, addressing token exhaustion challenges effectively.
The author shares their systematic approach to enhancing documentation quality in Notion by addressing duplication, outdated content, and implementing automated archiving processes.
Iterative refinement in AI workflows is crucial for enhancing productivity and developing user intuition, requiring responsive feedback and effective tracking mechanisms.
Triggers are essential for internal agents, enabling effective workflow initiation and iterative learning, with a focus on custom implementations over third-party tools like Zapier.
Including engineers in leadership meetings enhances decision-making and accountability, transforming management practices and fostering better communication within organizations.
Curiosity is essential in problem-solving, helping to reduce the costs of being wrong in decision-making and fostering better understanding in professional interactions.
Innovative hiring strategies from Stripe emphasize the importance of well-connected individuals, known as 'Lighthouse Hires,' to improve recruitment quality despite inherent challenges.
The post critiques the chaos of operating under pressure without a plan in tech projects and emphasizes the need for effective leadership to navigate such challenges.
Compound Engineering offers a structured approach to automate best practices in coding and project management, enhancing efficiency through AI tools.
Imprint shares insights on developing internal agent workflows, emphasizing practical learning and the importance of engaging with AI technologies for all companies.
A hybrid approach combining LLM-driven and code-driven workflows is essential for efficient software development, balancing innovation with reliability.
Agent Skills enhance internal workflows by improving context management and providing reusable snippets, as implemented by the engineering team at Imprint.
The author, a CTO with a background in software engineering, reflects on their return to coding after years of management. They discuss the challenges and benefits of writing software at work, emphasizing the importance of underst...
Internal adoption of LLMs relies on easy discovery, leading to the creation of a shared repository for Claude Code sessions using transcript extraction tools.
Subagents enhance internal agent workflows by improving context management, enabling concurrent tasks, and potentially offering security benefits, despite their current limited importance in practice.
Automating the merging of Dependabot pull requests on GitHub can streamline security updates, provided robust CI/CD practices are in place.
The author reviews significant personal and professional achievements over the past year, highlighting his new job, completed book, and evolving goals in engineering management.
The post showcases how to use Claude Code with Remotion to efficiently create visual summaries of blog data in under ten minutes.
The author recounts their transition from Mailchimp to Buttondown for newsletter management, highlighting frustrations with Mailchimp and a desire for greater ownership of their work.