# Agentic Coding

> **TL;DR.** Agentic coding is the software-development pattern where an autonomous AI agent plans, writes, tests, and iterates on code against a feedback loop — running its own builds, reading its own errors, and correcting without waiting for a human prompt between steps.

- **Category:** AI / Developer Tools
- **Stage:** established
- **Age:** 410 days
- **Origin date:** 2025-04-19
- **First detected:** 2026-04-16
- **Canonical URL:** https://earlyterms.com/term/agentic-coding
- **Sources:** 8 primary URLs

## Definition

Agentic coding is the software-development pattern where an autonomous AI agent plans, writes, tests, and iterates on code against a feedback loop — running its own builds, reading its own errors, and correcting without waiting for a human prompt between steps. It is explicitly contrasted with autocomplete-style "AI pair programming," where the human drives every token.

The term entered mainstream dev discourse with Anthropic's [Claude Code best-practices post](https://code.claude.com/docs/en/best-practices) (April 19, 2025, 614 points on HN) and Armin Ronacher's [Agentic Coding Recommendations](https://lucumr.pocoo.org/2025/6/12/agentic-coding/) (June 2025, 296 points). By April 2026 it had become a category name — Anthropic published a [2026 Agentic Coding Trends Report](https://resources.anthropic.com/2026-agentic-coding-trends-report) with Rakuten, CRED, TELUS, and Zapier case studies, and Qwen3.6-35B-A3B launched framed explicitly as "agentic coding power, now open to all" (1,262 HN points).

## Example

A developer writes "fix the flaky test in auth_test.go" — Claude Code reads the failing test, patches the assertion, runs `go test`, sees a second failure elsewhere, realizes a mock needs updating, fixes that, reruns the full suite, and opens a PR. No human keystrokes between task assignment and PR review.

## Analogy

The autocomplete era was spell-check for code; agentic coding is the difference between giving notes to a writer and hiring a contractor who ships the draft and checks it.

## Why it's emerging now

Three forces converged in April 2026: Anthropic published the first industry Agentic Coding Trends Report (Rakuten, CRED, TELUS, Zapier case studies), Qwen3.6-35B-A3B shipped as an open-weight model explicitly positioned for agentic coding (1,262 HN points), and the redesigned Claude Code desktop app shipped a parallel-agents UI. The term crossed from practitioner vocabulary to enterprise-buyer vocabulary.

## Related terms

- *related:* vibe coding
- *related:* claude-code
- *alias:* coding-agents
- *child:* parallel-agents
- *related:* agent-harness
- *related:* managed-agents
- *parent:* agentic AI
- *related:* Qwen3-Coder
- *competitor:* openai-codex-cli
- *competitor:* Cursor
- *competitor:* GitHub Copilot
- *related:* pair programming

## Sources

1. [Anthropic — Claude Code best practices for agentic coding](https://code.claude.com/docs/en/best-practices)
2. [Anthropic — 2026 Agentic Coding Trends Report](https://resources.anthropic.com/2026-agentic-coding-trends-report)
3. [Armin Ronacher — Agentic Coding Recommendations](https://lucumr.pocoo.org/2025/6/12/agentic-coding/)
4. [Martin Fowler — The role of developer skills in agentic coding](https://martinfowler.com/articles/exploring-gen-ai.html#memo-13)
5. [IBM — What is Agentic Coding?](https://www.ibm.com/think/topics/agentic-coding)
6. [Google Cloud — What is agentic coding?](https://cloud.google.com/discover/what-is-agentic-coding)
7. [Missing Semester (MIT) — Agentic Coding lecture](https://missing.csail.mit.edu/2026/agentic-coding/)
8. [Hacker News — Do you have any evidence agentic coding works?](https://news.ycombinator.com/item?id=42769919)

---
_Generated by EarlyTerms · https://earlyterms.com/term/agentic-coding_