Okakura Kakuzō, a curator at the Museum of Fine Arts, Boston, and a fixture in the salons of Isabella Stewart Gardner, spent his career helping Westerners bridge the gap toward Eastern aesthetics. In his lectures, he described a fundamental divergence in how cultures approach the “object” of art:
- The Western (Constructive) Perspective: Art is a conquest of matter. The artist is a creator who imposes their will and ego onto the medium to make something new. It is an additive process of construction.
- The Eastern (Revelatory) Perspective: Influenced by Taoism and Zen, art is an act of discovery. The artist “listens” to the material—a gnarled piece of wood or a blank scroll—to help it reveal its inherent spirit.
I found myself revisiting Okakura’s ideas during a 1-on-1 with a direct report. We were discussing a modern dilemma: how to adopt agentic coding as a standard engineering practice.
As it turns out, the “right” approach depends entirely on the state of your canvas.
Greenfields: The Western Approach (Creation)
When starting a project from scratch, the environment is a vacuum. To be effective, an AI agent requires a “conquest of matter”—you must actively build the world it inhabits. Ryan Lapopolo of OpenAI noted this in a recent post regarding their engineering work with Codex:
The lack of hands-on human coding introduced a different kind of engineering work, focused on systems, scaffolding, and leverage.
Early progress was slower than we expected, not because Codex was incapable, but because the environment was underspecified. The agent lacked the tools, abstractions, and internal structure required to make progress toward high-level goals. The primary job of our engineering team became enabling the agents to do useful work.
In practice, this meant working depth-first: breaking down larger goals into smaller building blocks (design, code, review, test, etc), prompting the agent to construct those blocks, and using them to unlock more complex tasks. When something failed, the fix was almost never “try harder.” Because the only way to make progress was to get Codex to do the work, human engineers always stepped into the task and asked: “what capability is missing, and how do we make it both legible and enforceable for the agent?”
In this phase, your work is additive. You are the architect, providing the scaffolding, tools, and abstractions that didn’t exist before. Without this “Western” drive to construct a framework, the agent has no ground to stand on.
Legacy Systems: The Eastern Approach (Revelation)
However, when you bring agentic coding into an existing codebase, the philosophy must shift. You are no longer filling a void; you are standing before a massive, weathered block of marble.
In a large, complex codebase, the “sculpture”—the clean, functional feature you want the agent to produce—already exists potentially within the code. The problem isn’t a lack of context; it’s an overabundance of noise. The agent is often overwhelmed by technical debt, circular dependencies, and “hallucination-inducing” abstractions.
Here, the engineering task is Eastern: your job is to reveal the path for the agent by removing context.
- Subtractive Engineering: You simplify interfaces, prune dead code, and decouple modules so the agent can “see” the intent of the logic.
- Protective Guardrails: Instead of adding “how-to” prompts, you build guardrails that prevent other engineers from re-introducing the “clutter” that obscured the sculpture in the first place.
As Okakura wrote in The Book of Tea:
“In leaving something unsaid the beholder is given a chance to complete the idea…”
In the world of agentic coding, if you refine the codebase enough to “leave the right things unsaid,” you give the agent the space to complete the masterpiece for you.
Deja un comentario