AI Generates 99.9% of My Code. The Engineering Work Didn’t Disappear

2026-07-13 · Source: AI on Medium · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning · Depth: Intermediate, long

Summary

A software development team has transitioned to an AI-assisted workflow where AI generates 90% to 99.9% of their code. The author, a Team Lead, details a three-year journey from basic ChatGPT use in 2023 to orchestrating multiple AI agents by 2026. This shift has moved engineering effort from manual coding to planning, context provision, validation, and system responsibility. Key developments included AI tools gaining access to IDE context in late August 2025 and dedicated desktop apps by February 2, 2026. The article emphasizes building a comprehensive "system" around AI tools, including `AGENTS.md`, Skills, and CLI integrations, to effectively onboard agents and manage the inherent "orchestration tax."

Key takeaway

For Team Leads overseeing software development, recognize that high AI code generation percentages do not equate to less work. Your role shifts to building and mentoring the AI-assisted development system, defining shared context, permanent rules, and validation processes. You must own the architectural decisions and ensure agents are onboarded with project-specific knowledge, matching validation effort to the blast radius of changes.

Key insights

High AI code generation shifts engineering focus to planning, context, validation, and system orchestration, not less work.

Principles

• Onboard agents like new developers.
• Plan before implementing meaningful changes.
• Treat "Recommended" as a proposal, not a directive.

Method

Establish an AI-assisted development system with `AGENTS.md`, Agent Skills, CLI tools, subagents, linters, tests, and human approval gates for planning and high-risk changes.

In practice

• Create `AGENTS.md` for repository rules.
• Develop Agent Skills for repeatable workflows.
• Use different models for planning vs. execution.

Topics

• AI-Assisted Development
• Software Engineering
• LLM Orchestration
• Developer Workflow
• AI Agents
• Code Generation
• Team Leadership

Best for: AI Engineer, Machine Learning Engineer, Director of AI/ML

Related on AIssential

• Paradox in action — Cheaper software production via AI increases demand for new applications, sustaining human software engineering roles.
• Slow down to speed up: so much has changed in 6 months’ time — Rapid AI agent adoption boosts individual dev output but risks software quality and organizational integrity without human oversight.
• Vibe Coding: The Complete Story (2015–2026) — The coding paradigm has shifted from human-driven to AI-agent-dominated, requiring new human oversight and expertise.
• How I use LLMs as a staff engineer in 2026 — Improved AI agents now enable engineers to offload significant coding, testing, and debugging tasks with light supervision.
• ‘Tokenmaxxing’ is making developers less productive than they think — Increased AI code generation correlates with disproportionately higher code churn and revision rates, undermining productivity claims.
• Who Understands the Code Now? — AI code agents boost speed but silently erode developer understanding, creating long-term maintenance risks.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI on Medium.