How I code with agents, without being 'technical'

2025-04-24 · Source: Ben's Bites · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Emerging Technologies & Innovation · Depth: Novice, long

Summary

An exited founder and investor details his "vibe-coding" approach, leveraging AI coding agents to build software without traditional programming skills. He describes spending 3 billion tokens in four months, exclusively using a terminal-based agent like Droid to develop various projects, including a personal site, a social tracker, a "wrapped" product for his company, custom CLIs, a crypto tracker, and an AI-directed video demo system. His workflow involves conversing with the model, entering "spec mode" to plan, linking documentation, and then letting the agent generate code, which he monitors and iterates upon. He emphasizes learning through identifying and resolving gaps in knowledge during the building process, utilizing an `agents.md` file for consistent setup and testing, and integrating agents with tools like GitHub and Telegram for mobile development.

Key takeaway

For entrepreneurs or product managers aiming to rapidly prototype and build software without deep coding expertise, you should adopt an agent-driven development approach. Focus on clearly articulating project specifications and iteratively refining agent outputs, treating bugs as learning opportunities rather than roadblocks. This method allows you to explore ideas quickly and contribute to products by understanding system architecture rather than mastering syntax.

Key insights

AI agents enable non-technical users to build complex software by abstracting traditional coding into a guided, iterative process.

Principles

• Embrace building ahead of capability to learn from failures.
• Prioritize terminal-based workflows for agent interaction.
• Customize agent instructions for simplified explanations.

Method

The author's method involves initiating a project with an AI agent via CLI, providing context and specifications, allowing the agent to generate code, and then iteratively testing, debugging, and refining the output while continuously learning from encountered issues.

In practice

• Use `agents.md` for consistent project setup and testing.
• Integrate agents with GitHub for PR reviews and fixes.
• Build custom CLIs for frequently used services.

Topics

• AI Coding Agents
• Non-Technical Development
• Command-Line Interfaces
• Agentic Workflow
• Software Development Paradigm

Code references

• jlongster/tigma
• clawdbot/clawdbot

Best for: AI Student, Software Engineer, Entrepreneur

Related on AIssential

• From journalist to app developer using Claude Code — Non-technical users can build production-grade software using a multi-agent AI workflow for planning, coding, and review.
• New course! Spec-Driven Development — Spec-driven development uses detailed specifications to guide AI agents, improving code quality and managing developer cognitive load.
• Why (Senior) Engineers Struggle to Build AI Agents — Philipp Schmid, Google DeepMind — Building AI agents demands a paradigm shift from deterministic software, embracing non-determinism, semantic understanding, and continuous evaluation.
• Open Source Coding Agents: The Future of AI-Driven Software Development — Open source coding agents enable autonomous, scalable software development by orchestrating multi-agent workflows and offering model flexibility.
• Jun 16, 2026Economic ResearchAgentic coding and persistent returns to expertise — Domain expertise, not coding skill, is the primary driver of success and agent utility in agentic coding.
• Coding Agents: AI Driven Dev Conference — The event offers practical, production-tested strategies for AI-assisted code development and agentic engineering.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Ben's Bites.