A tweet from Amir Salihefendic hit 288K views last week:
“Cursor feels like a company without product management. Just devs and designers running seemingly random experiments with little higher-level strategy. I use Cursor a lot, and it is becoming a very frustrating experience.” — @amix3k, Dec 18, 2025
He wasn’t alone. Two days earlier, kyzo’s tweet hit 482K views:
“guys at @cursor_ai can you please slow down with changing the ui every 2 days? i’m like constantly confused between the agent and editor, feels like the buttons change every day.” — @ky__zo, Dec 17, 2025
Dhruv Bhatia replied suggesting phased rollouts. I disagreed:
“Not even phased rollouts. Agile release train that thing. Every 6 weeks a big drop with 2 releases concurrent ala Chrome. Chug chug. Would make a huge impact.”
Someone asked if I was being sarcastic. I wasn’t. Chrome has done it 143 times. This is how platforms get built.
Release trains aren’t a shipping schedule. They’re an organizational forcing function.
They do two things continuous deployment doesn’t:
-
They reshape internal thinking. The question changes from “what can we ship today?” to “what can we ship in six weeks?” Conway’s Law works in reverse—cadence shapes scope ambition.
-
They tie strategy to market signals. Fixed date, variable scope. “We’re shipping Feb 1. What’s the story?” That conversation forces cohesion that reactive shipping can’t create.
Cursor is fast as hell. But fast isn’t the same as leading.
The internal effect
Organizations ship their org chart—that’s Conway’s Law. But it works the other direction too. Thaler and Sunstein call it choice architecture: the structure of decisions shapes the decisions themselves. A 6-week train changes the default question. Instead of “is this PR ready?” the question becomes “what’s our headline for this release?” Small stuff still ships—bug fixes, icon tweaks—but they’re not the organizing principle.
What about things that aren’t ready? Flag them off and merge anyway. Early adopters can flip the flags and play—that’s what Chrome Canary is. The train doesn’t mean everything ships to everyone. It means everything converges on a moment.
The external effect
A public cadence is a commitment. “We ship every 6 weeks” means someone has to answer: what’s going out? How do we tell the story? Big tech companies figured this out with annual conferences—Google I/O, WWDC, Autodesk University, Shopify Editions. The question becomes “what are you demoing on main stage this year?”
Smaller companies have their version: launch weeks. Supabase does this at a world-class level—they’re on Launch Week 15 now. I use them at Tilt and read every launch week post.
There’s a flip side too. Public releases create accountability in retrospect. “Hey team, let’s review your last six releases. What did you ship?” Steve Jobs had the Macintosh team sign the inside of the case—“real artists sign their work.” He wasn’t doing them a favor. He was putting their names on the record.
Standing ovations and shipping pressure
I worked at Autodesk from 2013-2018. We had an annual customer event, and during our SaaS conversion, “modern” devs initially scoffed at the yearly release cycle. But most users weren’t asking for faster releases. They wanted predictability and stability. We shipped plenty outside the primary window, but as an organizing principle, the cadence was insanely effective. I witnessed a standing ovation when the AutoCAD team demoed a feature that let people zoom and pan without losing their selection box. They could have shipped that sooner. But would it have had the same impact?
In 2018, I released something at Google I/O—one of the first large-scale WebAssembly applications, AutoCAD for the web. I was tech lead. I got to witness how everything laddered up to I/O. The focus it created was real.
Why Cursor specifically
Cursor has two problems trains would help. For acquisition: bigger releases create more noise, which compounds into impressions and top-of-funnel in a way that continuous small drops don’t. For retention: the kyzo tweet isn’t a product complaint—it’s a trust complaint. “Re-learning UI every day suuuucks” means users can’t build muscle memory. They can’t rely on you. Trains don’t answer what to build. They shape how.
But trains are boring. The metaphor is apt—you don’t want surprise improvements to a train schedule. Hot startups may resist boring on principle. If “ship as fast as fucking possible” is 90% of why Cursor retains great talent, then imposing cadence could kill the golden goose. I don’t know how much the chaos-energy matters to their culture. That’s the part I can’t see from outside. The obvious counter is “reactive speed is Cursor’s competitive advantage.” I don’t buy it—you just size the train for your market. Six weeks too long? Make it three. The only counter I’d actually accept: this is a champagne problem. Vocal minority on Twitter. Cursor is killing it on all fronts and there’s no actual issue.
One train at a time. Persistent, consistent. That’s how Chrome won the browser wars. Cursor could do the same.
Process note: Drafted from my tweets + notes. AI helped with structure and line edits. All claims are mine.