Midjourney Founder Flags AI Coding Tools Productivity and Drain
Midjourney founder David Holz said on July 7, 2026 that friends using the newest AI coding models feel simultaneously more productive and "extremely drained," according to a Business Insider report. The outlet says Holz's post on X drew responses from other practitioners, including ex-Meta engineer Shuming Hu, who said vibe coding "doesn't get you into a flow state," and Anthropic product lead Catherine Wu, who described preferring focused work with a single agent. For engineering teams, the exchange is an early signal that AI-assisted coding gains in raw output can carry a real cognitive-load cost, not just a productivity win. Several commenters suggested analog breaks and uninterrupted deep-work blocks as partial fixes; the account currently rests on this single report and has not been independently corroborated.
For teams scaling AI-assisted coding, the useful signal here isn't the raw productivity anecdote itself, it's that visible output gains and developer burnout can rise together. A one-off social post doesn't settle the question, but it puts a concrete, quotable frame around a tension many engineering leads are already managing privately.
What happened
According to Business Insider, Midjourney founder David Holz posted on X that his friends are "extremely productive and also extremely drained" while using the latest coding models. The outlet reported replies from other practitioners, including ex-Meta engineer Shuming Hu, who wrote that vibe coding "doesn't get you into a flow state," and Anthropic product lead Catherine Wu, who said she prefers focused work on a hard task with a single agent. Other commenters described analog breaks and uninterrupted, screen-free work as ways to offset the fatigue.
For practitioners
The pattern described lines up with well-documented limits on context switching and divided attention. Teams adopting AI coding tools should track cognitive load and burnout risk as a signal separate from commit volume or lines shipped, since short-term throughput gains can mask longer-term costs to code quality and retention. The remedies raised in the thread, single-agent focused sessions and scheduled disconnection from tooling, are low-cost experiments worth testing before a wider rollout.
What to watch
- •Developer-reported metrics beyond commit volume, such as time-to-merge, review depth, and self-reported fatigue.
- •Tooling or team policies that explicitly target attention fragmentation, such as single-agent, single-task workflows.
- •Additional public commentary or reporting, since this account currently rests on one outlet and has not yet been independently corroborated.
Key Points
- 1Midjourney's founder highlighted a pattern where AI-assisted coding lifts output while leaving some engineers reporting real mental fatigue.
- 2Practitioners in the thread favored single-agent focus sessions and offline breaks as low-cost ways to offset that fatigue.
- 3Teams scaling AI coding tools should track burnout signals like review depth and time-to-merge, not just commit volume.
Scoring Rationale
Adjusted down slightly from 5.9 to 5.0: this is a single social-media anecdote reported by one outlet about developer fatigue with AI coding tools, practitioner-relevant but not independently corroborated, a technical release, or a major event. Held above the visibility floor because the underlying operational pattern (output vs. cognitive load) is a genuine, topical signal for engineering teams.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems