Companies can use AI to code faster than ever before, but it doesn’t matter if the idea itself is terrible.
This is just one of veteran software engineer Dax Raad’s scathing critiques of the current state of AI in the workplace. His honest assessment resonates with many workers online.
Raad, the developer of OpenAuth, said the bottleneck facing companies is not coding productivity, but a lack of good ideas, unmotivated employees, corporate bureaucracy and “many other realities of shipping the real thing.”
Before AI, companies were held back by development costs, he argued in a Feb. 14 post about X, which also went viral on Reddit. Writing code has never been easier, but that doesn’t mean the original idea has any value. It also doesn’t mean that employees will simply use AI to write more code instead of working “with less energy.”
“Organizations rarely have good ideas, and it actually helped that ideas were expensive to implement,” Lard wrote.
Everyone talks about their teams as if they were at peak efficiency and their ability to write code was the bottleneck.
The actual situation is as follows
– Organizations rarely have good ideas. It actually helped that the idea was expensive to implement.
– The majority of workers…
— Dax (@thdxr) February 14, 2026
Describing the search for meaning among employees in the face of potential productivity gains from AI, Lard argued that “the vast majority of employees have no reason to be extremely motivated and just want to work a 9-to-5 job and get back to their old lives.”
Part of the problem, he wrote, is that some workers may simply use AI out of inertia. “They’re using this to mass-produce tasks using less energy,” he says. That can actually make the job more difficult for those who are doing their best, he wrote.
“The two people on the team who tried it were so fed up with the crappy code everyone was writing that they would quit right away,” Raad wrote.
There are also very real financial costs to equipping a company’s engineers with the AI coding tools and tokens needed to power new AI capabilities.
“What does it mean to your CFO that each engineer costs an extra $2,000 per month in LLM billing?” he wrote.
Raad, who is also the creator of OpenCode, an open source AI coding tool, said his criticism of AI should not be misunderstood.
“So many people don’t understand that I can criticize AI while building AI tools,” he wrote in another post about X.
OpenCode developers are beginning to develop a reputation for expressing skepticism about advances in AI, or at least the hype surrounding it. In a recent “Contrarian Talk,” Raad said his job is just as difficult as it was before AI.
“I’m tired of people feeling like all of a sudden the tables are turning and things are going to be easier,” he said during a talk on the AI Engineers podcast. “It’s not easy. My life is as hard as it’s ever been. Doing something great is as hard as it’s ever been. But that’s where all the fun comes from, that’s where all the purpose comes from.”
Raad’s views resonated within the development community. His post on X has been viewed about 793,000 times and received more than 22,000 upvotes from users who shared Lard’s thoughts on Reddit. On Medium, fellow engineer JP Capras praised Raad’s “honest opinion.”
“Ah, you all made me feel a lot less alone :D,” one Reddit user replied, adding that demands from management had “destroyed” the development process.
“It’s funny how accurate this is to my current situation,” reads the most-voted comment on the viral Reddit post.
Some companies say AI allows them to do more, faster, while others talk about the potential to “fail fast” and iterate. Okta, Salesforce, Snowflake, and Blackstone are all leaning heavily toward leveraging AI.
“It’s accelerated what’s possible,” Eric Kelleher, president and chief operating officer of digital identity company Okta, previously told Business Insider.
A new perspective on “AI fatigue”
Lard’s perspective comes amid widespread concerns about “AI fatigue.”
Siddhant Khare, who develops AI tools, recently wrote an essay describing how tools that were supposed to save us time ended up taking up our entire day. In short, the view is that even if AI improves productivity, it will come at the expense of a burnout workforce.
Steve Jaegge, a veteran of Amazon and Google, recently said companies should consider putting a three-hour cap on AI-assisted work or risk their workforces collapsing.
“I seriously think that engineering leaders at all levels, from founders to company leaders to line managers, need to be aware of this and realize that you may only get three hours of productive time from someone who is vibecoding at top speed,” Yegge told The Pragmatic Engineer newsletter.
Raad’s larger point is that AI is not a magic wand that companies can wave around and suddenly expect success, and the real problems remain.
“Even if we were to produce work faster, bureaucracy and many other realities of shipping the real thing would still be bottlenecks,” he wrote.
