Moving from Sculpting to Gardening: Rethinking AI Operations

Natalia Quintero argues that effective AI integration requires moving away from sculpting, where every detail is manually crafted, toward gardening, where the operator builds the conditions for systems to grow and self-correct. The implication is that the most powerful AI workflows are not those that automate tasks, but those that create feedback loops that improve over time. By shifting focus from building everything to managing systems, operators can achieve a competitive advantage that compounds quarterly. This approach helps executives and operators who feel overwhelmed by the vibe coding narrative and need a sustainable framework for scaling operations without drowning in technical debt.

The Hidden Cost of Vibe Coding Everything

The current excitement around AI often obscures a simple truth: just because you can build a custom solution for every problem does not mean you should. Quintero notes that while she successfully vibe coded a CRM using Google Sheets and AI, she eventually moved to a professional SaaS tool like Attio and Asana.

The rationale is counter-intuitive: professional software is not just a user interface; it is a compilation of thousands of logical rules gathered from years of industry experience. When you build your own tools from scratch, you are not just building an app; you are taking on the burden of maintaining those rules.

The question is should you build and maintain whatever you actually built? And I think in this case, and probably in other use cases, we also rolled out Asana for our project management system. I think we are able to do the scale of the work that we are able to do because of Attio, because of Asana, and because of Claudia managing all of that information is much greater than if we did not have those tools.

-- Natalia Quintero

The lesson is that effective systems leverage existing infrastructure to handle the bones of the operation, allowing the AI to act as the nervous system that coordinates workflows.

Why Immediate Pain Creates Lasting Moats

Most teams fail at AI integration because they try to remake the whole thing at once. Quintero’s success stems from the opposite approach: starting with the boring, standardized processes that already exist. By feeding existing OKRs and standard operating procedures into an AI agent like Claudie, she creates a system that can execute against a known standard.

The competitive advantage arises from the human sandwich, where the human provides the initial strategy and the final refinement, while the AI handles the execution in between. This creates a loop where the system learns from its own performance. Over time, this creates a moat: while competitors are struggling to manually track deals or triage emails, a system-oriented team has an AI that knows what is going on more than the human does.

The 18-Month Payoff: From Tasks to Systems

The most profound shift described is the move from viewing knowledge work as a series of individual tasks to viewing it as a system of loops. Quintero’s email triage system, which integrates with Asana and markdown files, is not just saving time; it is capturing context that would otherwise be lost.

Knowledge work now is turning into something like gardening, where when you are gardening You are creating the conditions for the growth to happen. But you are not like making the plant with your hands.

-- Natalia Quintero

When this logic is applied to personal life, such as the medical care portal Quintero built for her father, the downstream effect is a reduction in administrative friction that allows for higher-value human interaction. The payoff is not just efficiency; it is the ability to focus on being present and loving because the system has already handled the triage and follow-up.

Key Action Items

• Audit your existing SOPs: Before building an agent, document how you currently perform a single, repetitive task. If you cannot write it down, the AI cannot execute it. (Immediate)
• Identify the Bones vs. Brain: Stop trying to build your own CRM or project management system from scratch. Use established SaaS tools for the structural bones and use AI agents to act as the brain that inputs and retrieves data. (Immediate)
• Implement the Human Sandwich: For every automated workflow, ensure there is a clear human-led start (strategy) and end (quality control/taste) point. (Over the next quarter)
• Create a Second Brain for Context: Start routing your high-value communications into markdown files or centralized databases. This creates the data set that will allow your agents to make better decisions in 12 to 18 months. (Ongoing)
• Shift from Building to Gardening: Evaluate your current AI projects. Are you sculpting (manually fixing every output) or gardening (creating conditions for the system to improve)? If you are still sculpting, stop and refine your prompt or system logic. (12 to 18 months)