I write and ship code ~20–50x faster than I did 5 years ago

Two AIs, one coder: 50x faster claim sparks ‘is this even coding’ fight

TLDR: One dev claims 20–50x speed by using two AIs—one to build, one to review—and sticking to small, understandable changes. The crowd split: critics say it’s not “real coding” and doubt the 50x, while others want to copy the trick, debating IDE convenience and how to measure real impact.

A developer says he ships code up to 50x faster by running two AI chatbots in browser tabs — one “builder” to write code with full context and one “reviewer” to catch mistakes — and the comments section went full reality TV. He keeps changes surgical, pastes whole files for context, has the AI explain the plan first, and uses the second AI to review diffs before shipping. No fancy plugins, just tabs and a terminal, about $40/month, and a strict rule: never accept code you don’t understand. There’s even a playbook.

Cue the brawl. The top clapback: “You aren’t writing code — the machine is.” Skeptics piled on the math: “50x what?” asks one, demanding real metrics (lines of code? merged pull requests?). Another snarked it’s just boilerplate at light speed, not original thinking. Practical folks asked how to paste diffs and whether an in-editor assistant would be less clunky. Meanwhile, a quieter crew said “thanks” and vowed to try a second AI to review git diff. The meme energy? Think angel AI vs devil AI on your shoulders and “two interns who never sleep.” Love it or loathe it, this thread turned a workflow tip into a culture war over what it means to “code” in 2026.

Key Points

  • The author uses two browser-based AIs in parallel: one as a builder with full context and one as a reviewer for diffs.
  • Full-context inputs (entire files/modules) and upfront discussion of approaches and tradeoffs keep architectural control with the developer.
  • A defined workflow implements changes across files, then uses a second AI to detect regressions and subtle issues in diffs.
  • The approach accelerates cross-stack development across languages (e.g., Swift, Objective-C, JavaScript) and boundaries.
  • The method emphasizes surgical edits, small reviewable diffs, and discipline; reported cost is about $40/month.

Hottest takes

"You are getting a machine to do it." — chrisjj
"It’s hard to believe all these 50x AI productivity claims" — eloisius
"I might start using a second LLM to review the diffs." — hu3
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.