AI-first company

Company case

Shopify: an AI-first memo, over half the code from AI by its leaders’ count, and unread AI work passed to colleagues

In April 2025 Shopify’s chief executive made AI use a baseline expectation for every employee, and told teams to show why AI could not do the work before asking for more staff. Since then Shopify’s headcount has fallen slightly while revenue rose 30%, and its leaders say AI writes over half its code. They also name what went wrong: a leaderboard of AI use that rewarded the wrong thing, more code than reviewers can keep up with, and AI work passed to colleagues unread. Every figure is the company’s own.

By
Last reviewed
Corrections
Report an error
Outcome Company-reported result

The company published the figures itself.

At a glance

Company
ShopifyCommerce software for online and in-store sellers
Size
Enterprise (1,000+ employees) Shopify reported about 8,100 employees at 31 December 2024 and about 7,600 at 31 December 2025 in its annual reports.
Industry
E-commerce and retail Shopify’s customers are online retailers. The case is about how Shopify’s own staff use AI, not about its shops.
Where and when
Canada, with staff working remotely in many countries. 2025–2026.
Who produced the figures
The company itselfEvery figure on this page comes from Shopify: its annual reports filed with the US Securities and Exchange Commission, its earnings call, its engineering blog and its executives’ interviews. The shares of work done by AI are executives’ estimates. No one outside the company has measured them.

What they did

On 7 April 2025 Shopify’s chief executive, Tobi Lütke, posted an internal memo on X. He wrote that he was sharing it because it “was in the process of being leaked”. Its first line: “Reflexive AI usage is now a baseline expectation at Shopify”. Four of its rules:

  • “Using AI effectively is now a fundamental expectation of everyone at Shopify.”
  • “We will add AI usage questions to our performance and peer review questionnaire.”
  • “Before asking for more Headcount and resources, teams must demonstrate why they cannot get what they want done using AI.”
  • “Everyone means everyone. This applies to all of us—including me and the executive team.”

The memo also said AI must be part of the prototype phase of every project. It pointed staff to an internal AI chat tool, and developers to tools that were “all pre-tooled and ready to go”: Copilot, Cursor, Claude Code and a shared gateway to AI models. Learning was “self directed”, with staff asked to share what they learned. The memo set no target for headcount and said nothing about layoffs.

The numbers

From Shopify’s annual reports:

  • About 8,100 employees at the end of 2024 and about 7,600 at the end of 2025.
  • Revenue of $11,556 million in 2025, up 30% from $8,880 million in 2024.
  • The reports do not link the fall in headcount to AI. Shopify explains its higher research and development costs by “increases of $105 million in employee-related costs and $61 million in computer hardware and software costs”.

Between November 2025 and April 2026 Shopify also cut jobs in several small rounds. It described each as a change of focus or structure, and gave no numbers.

From Shopify’s executives, on the earnings call for the first quarter of 2026 and in interviews:

  • More products, same headcount. The president, Harley Finkelstein: “We shipped over 300 new products and features last year alone. We kept our flat headcount”.
  • Share of code written by AI. Lütke said on 4 May 2026 that AI writes “over 50%” of Shopify’s code. On the earnings call the next day, Finkelstein said “AI right now writes well over 50% of our code today”.
  • One internal AI agent. On 28 May 2026 Shopify’s engineering blog reported that River, an AI agent staff ask for work in the company chat, coauthored “one in eight merged pull requests across Shopify”: 3,536 in 30 days. A pull request is a proposed change to the code that a colleague checks before it goes in. On 15 September 2026 Lütke gave a much higher figure: “I want to say it’s probably up to about 50% of pull requests in Shopify” now come from conversations in the company chat rather than from engineers writing code themselves.
  • No limit on AI spending. The chief technology officer, Mikhail Parakhin: “we effectively fund unlimited tokens for everybody”. Tokens are the units AI use is paid in. According to Bessemer Venture Partners, an alert goes out when someone spends more than $250 a day, and is followed up rather than blocked.
  • Productivity. The head of engineering, Farhan Thawar, put the gain in engineering at roughly 20%, which he called “a humble estimate”. He gave no method.
  • More interns. Shopify took about 1,000 interns in 2025 and said it would take 1,000 again in 2026.

The AI figures measure different things, at different dates. “Over 50%” of code is two executives’ estimate for all AI tools. One in eight pull requests is the company blog’s count for one agent over 30 days. “Up to about 50%” of pull requests is the chief executive’s estimate four months later. Shopify has published no method for any of them, and no one outside the company has checked them. Nothing in the filings separates the effect of AI from the job cuts.

What happened next

  • The chief executive calls it a success. In May 2026 Lütke said of the change: “I think this worked extremely well.” Of the memo itself: “It was probably slightly too early to write it, but if you read it now, it says nothing that would be surprising.” Shopify has not withdrawn the memo. In April 2026 Bessemer Venture Partners wrote that staff were still assessed in their twice-yearly performance reviews on how “AI-reflexive” they are.
  • More interns, not fewer. The memo asks teams to prove AI cannot do the work before they hire. Shopify still raised its intern intake to about 1,000 a year. Lütke: “We take 1,000 interns a year.” Its internship site says it hires more interns “because we have gone all in on AI”.

What Shopify’s leaders say went wrong

  • A leaderboard of AI spending. Shopify first ranked staff by how many tokens they used. In 2025 Thawar defended the ranking: “I know folks who are proud to show up on the top 10 token spend leaderboard due to valuable work being done”. In May 2026 Lütke said: “At some point we had a leaderboard and so on of token consumption. Of course, that leads immediately to bad effects, so we don’t have that.” In June 2026 Thawar gave the reason: “people wanted to be on the leaderboard, and of course the incentives were, like, misaligned”. Each person now sees their own use, and how it ranks in their department, on their internal profile. Parakhin: “it’s not about just consuming tokens”.
  • More code than reviewers can handle. In April 2026 Parakhin said: “Lines of codes are exploding for everybody right now … you have to have a very strong narrow waist during PR review. Otherwise, just the number of bugs will go through the roof.” He said a good model now writes code with fewer bugs on average than the average person, “But since they write so much more of it, like more of it will make it into production.” He described build-and-test pipelines across the industry as “creaking”, with “the number of bugs slipping by increasing”. He spoke about “everybody”, Shopify included, and gave no Shopify figure for bugs. Shopify built its own review system using expensive AI models. Thawar, the same month: “We still require a human PR reviewer, which is now becoming a big bottleneck because if lots of code is being generated by AI, more time is needed to review the code”.
  • Unread AI work passed to colleagues. In September 2026 Lütke was asked whether AI had made anything worse at Shopify. He said: “One thing that’s definitely worse is the failure case now of lazy work is not lack of output, it’s actually over output”. And: “Internally, we have come to call these things that people are lobbing ‘slop grenades’ at each other”. He described how it happens: “You just tell the AI to go nuts. It makes a pull request. You just say, ‘Yep, that’s good.’ You don’t really read it, and now it has to be reviewed by your colleagues.” Long emails blown up by AI fall under the same name: “that’s definitely a bad thing.”
  • A changed view on new graduates. In May 2026 Lütke said he had expected engineers straight out of university, used to AI from the start, to have a big advantage. “I had to change my mind here”. He found that experienced engineers steer the AI tools well enough to do a great deal in very little time.

What the case shows

  • A memo spreads AI use. It does not make the work good. Lütke says use “skyrocketed the second this message came along”. Seventeen months after the memo, he named unread AI work as the thing AI had made “definitely worse”.
  • Do not reward AI use by volume. A ranking of token spending rewarded spending. Shopify kept unlimited spending, but replaced the ranking with a private view of each person’s use.
  • Plan the checking before you add more AI output. At Shopify, more code from AI meant more for people to review, and the company’s own chief technology officer warned about bugs reaching production.
  • The person who asks the AI owns the result. Lütke’s complaint is about people who approve AI work they have not read. Thawar’s rule is that “The human still has to be responsible of what goes to production”.
  • Separate the firm figures from the estimates. Headcount and revenue come from audited annual reports. The shares of code and pull requests written by AI are executives’ estimates, and they do not agree with each other.

Shopify and Fiverr both used the “AI-first” banner. Fiverr cut about 30% of its staff in one go. Shopify’s headcount fell by about 500 over 2025 while it took on more interns. The same month as Shopify’s memo, Duolingo’s chief executive sent a similar one, which came with a change to one process: writing courses.

What this says about the practices

Practices the company applied

Practices that were missing

What to take from it

If someone cites Shopify as proof that AI pays off, keep in mind that the company published the figures itself. Treat the case as an example of what one company did, not as a forecast of what AI will do for you. A practice that was missing here, to check in your own plans: Budget for the work AI creates.

Sources

  1. Tobi Lütke (7 April 2025). “Reflexive AI usage is now a baseline expectation at Shopify.” Internal memo, posted by Shopify’s chief executive on X. x.com
  2. Shopify Inc. (11 February 2026). Annual report for 2025 (Form 10-K), filed with the US Securities and Exchange Commission. Source of the 2025 employee count, revenue and research and development costs. sec.gov
  3. Shopify Inc. (11 February 2025). Annual report for 2024 (Form 10-K), filed with the US Securities and Exchange Commission. Source of the 2024 employee count. sec.gov
  4. Shopify (5 May 2026). First-quarter 2026 earnings call. Transcript by The Motley Fool. fool.com
  5. Shopify Engineering (28 May 2026). Under the River. Shopify’s engineering blog. shopify.engineering
  6. Uncapped with Jack Altman, episode 50 (20 May 2026). Tobi Lütke – Building Shopify and the Future of AI. Transcript on the show’s site. uncappedpod.com
  7. Latent Space (22 April 2026). Shopify’s AI Phase Transition, with Mikhail Parakhin, Shopify CTO. Transcript on the show’s site. latent.space
  8. The Knowledge Project with Shane Parrish (15 September 2026). Tobi Lütke: AI Agents, Better Decisions, and the Future of Work. Transcript on the show’s site. fs.blog
  9. 20VC with Harry Stebbings (4 May 2026). Shopify’s Tobi Lütke on How AI is a Scapegoat for Mass Layoffs & What Will Labour Markets Be in the Future. Episode page. podcasts.apple.com
  10. Bessemer Venture Partners (2 April 2026). Inside Shopify’s AI-first engineering playbook. A venture capital firm’s write-up of an interview with Shopify’s head of engineering, Farhan Thawar. bvp.com
  11. VentureBeat, Beyond the Pilot, season 1 episode 16 (24 June 2026). Small Models, Massive Wins: The New Shopify AI Formula. Interview with Farhan Thawar, with transcript. venturebeat.com
  12. First Round Review (15 July 2025). From Memo to Movement: Shopify’s Cultural Adoption of AI. A venture capital firm’s publication, interviewing Shopify executives. review.firstround.com
  13. Shopify Internships. Zero experience, infinite leverage. Shopify’s internship site. internships.shopify.com
  14. BetaKit (22 January 2026). Shopify makes more job cuts, this time targeting partnerships division. betakit.com
  15. The Logic (4 May 2026). Shopify makes cuts to its operations team in latest round of layoffs. thelogic.co