Agentic EngineeringPractical application of AI to create quality software

Software development is changing and we find ourselves at a convergence. Between the extremes of technological zealotry ("all code will be AI-generated") and dismissive skepticism ("AI-generated code is garbage") lies a more practical and nuanced approach—one that is ours to discover together.

Agentic Engineering

Combining human craftsmanship with AI tools to build better software.

Agentic Engineering means integrating AI into your existing development workflow. When quality software is the goal, there is no substitute for a skilled engineer. It is about enhancing what we can accomplish through thoughtful collaboration.

Perfecting our tools has always been essential to good engineering. Traditional developer tools are predictable—they give the same output for the same input. AI tools work differently—they're powerful but stochastic.

Learning to work effectively with this inconsistency is a new engineering skill.

Leverage, not magic

Agentic Engineering represents the integration of two seemingly opposing approaches:

[ Fig 1 ] — Traditional

Fast, reliable, deterministic development with predictable outcomes.

[ Fig 2 ] — Emerging

Fluid, stochastic, exploratory potential with creative possibilities.

[ Fig 3 ] — The New World

Interwoven relationship between the predictable & unpredictable.

Core principles

Quality is our responsibility

As engineers, we are solely responsible for the quality of what we build. It's up to us to develop judgment about when AI improves our outcomes and when it doesn't. There's no prescribed formula (yet); we have to build our own understanding of how AI best fits into our craft.

Craftsmanship is critical

Quality software is the goal. AI gives us new leverage, but doesn't replace the need for taste and judgment. Instead, it amplifies the impact of engineering expertise by handling repetitive tasks, letting us focus on what matters: creating software that's not just built faster, but built better.

The Case for Software Craftsmanship in the Era of Vibes

Read Post →

A post by

Nathan Sobo

Skills improve outcomes

Working with agents to write code requires rigor, clarity, and intent. Learning to effectively direct AI tools is a skill that demands practice and refinement. Just as we've learned to master other tools in our stack, we need to develop expertise in working with AI agents.

A new way to collaborate

Traditional development tools were built for human-to-human workflows, not real-time AI interaction. Today's workflows demand tools that support rapid feedback loops, parallel agent conversations, and efficient review of AI-suggested changes—all while maintaining code clarity and quality.

Let's learn together

AI agents have not existed for very long. We've already learned a lot about how best to use them, but there remains much to learn. What's more, the tools themselves continue to improve, which creates ever-moving goalposts in our pursuit of understanding.

In this rapidly evolving environment, how can we determine which skills and techniques work best with which tools? The same way we always have: sharing knowledge with one another!