Seattle software engineers say Cursor, an AI-powered code editor, is revolutionizing how they work — boosting productivity, enabling faster prototyping, and opening the door to more ambitious projects.
Pioneer Square Labs hosted the first-ever Cursor Seattle meetup on Tuesday, reflecting the rising popularity of fast-growing AI coding tools being adopted by programmers.
Cursor integrates large language models from OpenAI, Anthropic, and others to suggest, write, and refactor code, aiming to make development faster and smarter.
“It’s a massive productivity boost,” said Gabe Brown, co-founder of Big Box, a Seattle video game startup acquired by Meta in 2021. Brown recently “vibe-coded” an app in 90 minutes — something he said would have taken four weeks without the help of Cursor.
Matt Deitke, co-founder of Seattle startup Vercept, told GeekWire it’s “fundamentally changing how much code you can write.”
The buzz isn’t limited to startups. Amazon is reportedly in talks to deploy Cursor internally — even as the Seattle tech giant builds its own AI coding assistant — in a sign of growing interest from tech giants.
Investors are also taking notice.
Anysphere, the small San Francisco startup behind Cursor, recently hit a $9.9 billion valuation after raising $900 million — its third fundraise in less than a year. The startup was founded in 2022.
Windsurf, a rival to Cursor, agreed last month to be acquired by OpenAI for about $3 billion.
Cursor also competes with GitHub Copilot, Replit, and others.
Kevin Leneway and Jared Kofron, engineers at Seattle startup studio Pioneer Square Labs, kicked off the event on Tuesday by using Cursor to build an icebreaker game in under 30 minutes. The end result wasn’t quite perfect but did generate a working demo.
Nick Miller, an engineer with Cursor, later joined via video to share insights about the tool. Miller said 95% of his code is now written by a machine.
The software engineer’s job description is starting to change, Miller said.
“I’m really more supervising and orchestrating, as opposed to writing code by hand,” he said. “I think we’re going in that direction.”
Leneway, who posts Cursor tutorial videos on YouTube, echoed that sentiment.
“My daily job has changed from writing code to acting more as a project manager to define what I want built, then reviewing code the Cursor agent wrote,” he told GeekWire. “I personally have a lot more job satisfaction now since I can work at a bit of a more strategic level focused on architecture and product rather than typing out code and manually tracking down bugs.”
Like many nascent AI tools, Cursor offers powerful advantages but also has limitations. Some users complain that Cursor adds bugs or makes their code worse. And hallucination is still a concern. Cursor itself sparked headlines last month after its AI support bot generated a company policy that doesn’t actually exist.
Miller advised engineers to focus on using Cursor’s “Rules” feature and attaching documentation to help guide the coding agents. He also stressed creating new “context windows” for individual projects or tasks.
“The key is sort of ‘context window engineering’ — thinking about when to start new session, thinking about how to start those sessions, and really being aware of what is in that context window,” Miller said.
AI coding tools are also sparking questions about whether AI will replace software engineering jobs.
But inside the meetup on Tuesday in Seattle, the feeling was more of excitement than fear.
Harshitha Rebala, who recently graduated from the University of Washington and just joined Vercept, said she’s using Cursor to get questions answered.
“It’s really helpful to get ramped up on the codebase,” she said.
Ron Theis, a longtime software engineer in Seattle, said he’s now doing much more code review and a “higher level of problem solving.”
“The number of curly braces I’ve typed since I started using Cursor has gone way down,” Theis said, alluding to the common punctuation used in programming.
Read the full article here
