Why companies are really going AI-first

Estimated reading time: 6 minutes

Tech companies like Duolingo, Klarna, and Shopify are putting AI at the heart of their operations — redefining not only how the work gets done, but also by who — human or AI?

Across the tech industry, AI is no longer just augmenting work – it’s actively reshaping teams. In recent months, high-profile companies have begun codifying an AI-first hiring philosophy, signaling a shift that engineering leaders can’t afford to ignore. 

From Duolingo to Shopify and Klarna, executives are drawing a clear line: roles must now justify their resistance to automation. 

In May 2025, Duolingo’s Chief Executive, Luis von Ahn, announced in an internal email, later shared on LinkedIn, that the company would “gradually stop using contractors to do work that AI can handle” and only hire “for roles that cannot be automated.” 

In April 2025, an internal leaked memo, which was posted later on X by Shopify’s Chief Executive Tobi Lütke, declared that teams had to “prove why certain jobs can’t be done using AI” before greenlighting any new roles.

Following in this trend, Klarna’s Chief Executive, Sebastian Siemiatkowski, recently revealed the company reduced its headcount by roughly 40% as a result of AI, shrinking from 5,527 employees at the end of 2022 to 3,422 a year later. 

He attributed the cuts to a combination of natural attrition and greater use of AI. “The truth is, the company has shrunk from about 5,000 to now almost 3,000 employees,” he told CNBC.

The human toll is becoming harder to ignore. “This AI-first approach is basically corporate speak for mass layoffs,” one Reddit user commented in response to Duolingo’s announcement. 

Another user, referencing Shopify, added: “I can’t imagine any developer now or in the future thinking their job at this company would have any security to it.”

What’s really going on?

As AI continues to reshape the tech industry, companies are being forced to respond. Gergely Orosz wrote on Bluesky, “when you see a big wave: the harm of being too optimistic is less than being too pessimistic.”

This is because companies fear that if they do not jump on the bandwagon and implement AI-first strategies, they risk being outperformed by competitors who have. 

In the memo, Lütke declared that “using AI effectively is now a fundamental expectation of everyone at Shopify.” He also made it clear that this policy didn’t just apply to engineering. “Everyone means everyone.”

Following suit in May 2025, Duolingo’s Chief Executive, Luis von Ahn, announced in an internal email, later shared on LinkedIn, that the company is going “AI-first.”

“Being AI-first means we will need to rethink much of how we work,” he wrote. “Making minor tweaks to systems designed for humans won’t get us there.” 

He added that the company will have to start from scratch and that it won’t rebuild overnight, especially as “getting AI to understand [Duolingo’s} database will take time.”

However, Ahn stressed that the company can not wait until the technology is 100% and has to “move with urgency” rather than “moving slowly and missing the moment.”

Duolingo calls this “a fundamental cultural shift” where departments must “reconsider their workflows.”

Anxiety rising

However, the AI-first shift in companies has caused growing anxiety across the industry. This fear is heightened in the context of the mass industry layoffs that started in 2022 and have resulted in over 600,000+ tech workers losing their jobs since, according to layoffs.fyi. 

A recent Clarify Capital survey also found that American tech workers are more worried about job security than those in any other industry, with one in three fearing layoffs in 2025.

“There’s been a deliberate effort by big tech to ‘put workers in their place.’ Framing coders as replaceable [with AI] undermines their value and reinforces that message,” says Dash.

In the blog post titled The Dumbest Move in Tech Right Now: Laying Off Developers Because of AI, Paolo Perazzo argues that AI should be used to empower developers – not replace them. “This isn’t a time to replace teams with AI,” he writes. “It’s time to invest in them, to empower them.”

“Laying off developers is a management and vision issue more than a reflection of AI advancement.” Perazzo adds. 

He argues that a leader’s mission is not to maintain current output with a reduced headcount but to leverage AI to deliver 10X value with existing teams. 

“The question isn’t ‘How many developers can we cut?’ but rather, ‘What can we finally build now that the constraints have been lifted?’” He adds. 

10x yourself

Despite Klarna’s Siemiatkowski saying “we’re going to stop hiring,” Klarna has continued to post open roles, primarily in Europe.

Meanwhile, Von Ahn argues that the firm’s AI-first strategy isn’t about replacing staff with AI. Instead, he says that the changes are “about removing bottlenecks” so that employees can “focus on creative work and real problems, not repetitive tasks.”

“We are all lucky to work with some amazing colleagues, the kind who contribute 10X of what was previously thought possible,” Lütke wrote. “And what’s even more amazing is that, for the first time, we see the tools become 10X themselves.”

While these are extreme examples by a handful of vocal CEOs at venture-backed companies, it does underscore the growing demand for engineers to work on their AI proficiency.

A Forbes survey found that 71% of employers now prioritize AI skills over traditional experience when hiring.

“It is clear that we have to relearn how we do things in our day-to-day work. Productivity expectations are rising, and being AI-proficient is becoming essential. There is a real risk that those who don’t keep up will fall behind,” says Ashwin Das Gururaja, senior engineering manager at Adobe.

“It’s still relatively early for AI to have its full impact on hiring, but we’re certainly seeing an expectation of familiarity in the overall AI landscape, and a basic fluency in the most common AI platforms,” says Dash. 

A growing performance divide 

Ashwin warns of a growing performance divide between those who are willing to leverage AI tools and those who are more resistant. 

For example, a study of developers using GitHub Copilot found they completed coding tasks 55% faster than those without AI assistance, while a large-scale trial at Google showed a 21% reduction in task time among engineers using AI tools. 

Ashwin explains that engineers who adopt AI-first tools like Cursor and Windsurf are already seeing notable productivity gains – prototyping faster, reducing repetitive work, and iterating more effectively.

However, he warns that even a short delay in adopting AI tools could soon lead to “large performance gaps” between those ahead of the curve and those falling behind. 

“A small group of top performers using AI tools will deliver output equivalent to what the entire organization produced previously,” Ashwin argues.

AI performance reviews 

Certain organizations seem all too keen to capitalize on these AI productivity gains, pushing engineers to showcase how they use AI in their work as part of their performance reviews.  

Shopify and Duolingo have noted in their internal memos that AI usage questions would be added to performance and peer review questionnaires.

AI use would also be “part of what we evaluate in performance reviews,” von Ahn explained. 

Lütke also said that using “AI effectively” would be measured in performance reviews.

Although using AI effectively is becoming a consideration in performance reviews, Dash argues that it is hard to imagine how this could be measured due to LLMs’ tendency to hallucinate and produce inconsistent results. 

“This seems like an area that will be subject to extremely subjective evaluations, and therefore may fall prey to biased or skewed judgments of worker effectiveness, since there are very few objective measures of how well a worker is incorporating AI into their work streams. 

“Crude measures like ‘how often did you ask the AI for help’ will end up with meaningless responses, but may be the best that companies can do to start.”

Get articles like this in your inbox

Choose your LeadDev newsletters to subscribe to.