Back to Blog
High Level for Humans, Low Level for Agents: Rethinking Development in the AI Era
AI Strategy

High Level for Humans, Low Level for Agents: Rethinking Development in the AI Era

Max Li
Max Li
March 10, 2026

I’m the founder of Grassrootech, and I also teach Mobile Application Development (MAD) at Merrimack College. When I first taught the course two years ago, the development environment looked very different from today.

Back in Spring 2024, Google’s Android Studio had only primitive AI integration. Fast forward to now, and Google has embedded Gemini across many of its products. Android Studio is no exception. When developing Android apps today, developers can use Gemini to generate code, suggest fixes, and even help design parts of an application.

Seeing how quickly AI has become part of the development workflow, I decided to incorporate it into a midterm exam question for my students.

Given the fact that AI can do coding for us, which is more important for students to grasp: (1) high-level concepts or (2) low-level details? Justify your answer with concrete examples.

It’s clearly a subjective question, but the expected answer leans toward high-level concepts.

The Abstraction Shift

Why? Because interacting with AI is fundamentally about abstraction. The prompts we write are essentially high-level descriptions of goals. We tell the AI what we want to achieve, and the AI fills in the implementation details. In many ways, AI agents operate by breaking a high-level goal into smaller sub-goals, solving each piece, and then combining the results to produce the final outcome.

This shift changes the role of developers. Humans increasingly focus on defining the problem, structuring the solution, and validating the results. The AI handles much of the detailed implementation.

You can think of it like a company structure. The boss defines the vision and the goals. The team works through the detailed tasks needed to achieve them.

High level for humans. Low level for agents.

Max Li

Max Li

Founder, Grassrootech

max@grassrootech.com

Max is dedicated to bridging the gap between advanced research and practical industry application. Drawing on his experience at IBM Research and Union University, he leads the development of AI solutions that drive meaningful progress.