GitLab Act 2 — The End of CREDIT and a Handbook Company Abandoning Its Identity

Is GitLab’s announcement of CREDIT’s end a simple refresh of corporate values, or a formal renunciation of the “all-remote, transparent, handbook-first” Silicon Valley antithesis identity? And if it is the latter, what exactly have we been calling “the model of a transparent company” for the past 12 years?

Introduction — The Opening Scene of the Second Act

In the early hours of May 4, 2026, GitLab CEO Bill Staples published a post titled “GitLab Act 2” on the company’s official blog. At the same time, layoff notices went out internally. GitLab will end operations in up to 30% of the 18 countries where it has been active, flatten management hierarchies in some departments by as many as three layers, and reorganize R&D into roughly 60 smaller, more independent teams. A voluntary-departure application window remains open until May 18. In the same announcement, the company also signaled the end of CREDIT — Collaboration, Results, Efficiency, Diversity·Inclusion·Belonging, Iteration, Transparency — the values that have been called the company’s soul for 12 years. In their place stand three new operating principles: Speed with Quality, Ownership Mindset, and Customer Outcomes.

The moment the announcement hit the Hacker News front page, two reactions appeared in the thread at the same time. One was a shrug: “a company just announced a fairly ordinary downsizing.” The other was indignation: “this isn’t a downsizing, it’s a declaration of identity abandonment.” Both are partly right. The absolute scale of the headcount cut is small compared with other Big Tech examples. But because what GitLab declared it was abandoning was not a personnel policy but the company’s reason for being, this announcement carries a different weight from an ordinary layoff notice.

GitLab began in Ukraine in 2014 as an office-less company from day one. Its CEO and CTO lived on different continents; every meeting minute, every HR policy, and every compensation structure was written into a public handbook. That handbook eventually exceeded 2,700 pages and became the de facto standard text for “how a transparent remote company is run.” The D — Diversity, Inclusion, Belonging — and T — Transparency — of CREDIT were the two pillars of that identity. In the early hours of May 4, GitLab decided not to name those two pillars anywhere in its new operating principles. If GitHub a week earlier appeared to be cracking along five axes simultaneously, GitLab in a single announcement set down both pillars of its identity at once. What is happening to the Git forge SaaS category itself?

Section 1 — What Exactly Did Act 2 Announce

Let us start with the announcement itself. At the top of the post, Bill Staples defined the restructuring as “reshaping the company for the agentic era.” His diagnosis can be summarized in two lines: “Operationally, we grew into a shape that was right for the last era and isn’t right for this one.” And: “The agentic era affords GitLab the largest opportunity in our history as a company.” The fact that the phrase “largest opportunity” appears inside a layoff announcement already tells you something about the tone of the post.

The concrete structural changes fall into four lines. First, country footprint reduction. GitLab will close operations in up to 30% of the 18 countries where it currently maintains formal employment infrastructure. As Simon Willison noted in his commentary, this means roughly five countries listed on the handbook’s employment-solutions page will disappear. Second, flattening management layers. In some departments, up to three layers of management will be removed. Third, R&D reorganization. Existing teams will be reshaped into about 60 smaller, more autonomous teams, nearly doubling the number of independent teams. Fourth, the voluntary-departure window. Applications will be accepted until May 18, with a corresponding package. The cost savings, the post says, will be reinvested into accelerating R&D and strengthening GTM (Go-To-Market). Q1 and FY27 guidance remain unchanged; financial details will be shared on the June 2 earnings call.

The end of CREDIT is treated as a separate chapter within the same announcement. Staples acknowledged that CREDIT was “the set of values that led the company through its first chapter,” but argued that new operating principles are needed for the second chapter. The three principles he laid out are: Speed with Quality — moving faster, with stronger guardrails. Ownership Mindset — autonomous decision-making by those closest to the work. Customer Outcomes — measuring success by customer impact rather than internal activity. Read purely as English phrases, the three items are very ordinary OKR headers for a SaaS company. But because what disappeared from that lineup are D and T, the ordinariness inverts in meaning.

Let us map which of the six letters of CREDIT were absorbed into the new principles and which were dropped. C (Collaboration) can be seen as distributed inside the autonomous decision-making of Ownership Mindset. R (Results) and E (Efficiency) were absorbed into Speed with Quality and Customer Outcomes. I (Iteration) fits into the “faster” part of Speed with Quality. That leaves D — Diversity·Inclusion·Belonging — and T — Transparency. Where the two letters went is never explicitly answered in the post. Neither the names of the new operating principles nor the body text contain the words “diversity” or “transparency.” It is hard to see as a coincidence that a company unveiling new values failed to specify which old ones it discarded.

Another point worth noting is this line, written by Staples himself: “Our transparent restructure process creates uncertainty that is real and it’s hard, and I’m not going to pretend otherwise.” HN user hemul3n quipped that this “reads suspiciously like Claude.” The more substantive issue lies elsewhere. Removing T from the values list while using the phrase “transparent restructure process” in the same post is not so much a contradiction as a signal that T has been demoted from a value to a technique. A word that is dropped from values and survives only as an adjective vanishes from the company’s identity but lingers in marketing copy.

Section 2 — Why Is a Handbook Company Changing Its Identity Now

The real reason GitLab swapped out its values is not written in the announcement itself. But there are plenty of clues to trace it.

The most direct clue is the stock price. As simonw pointed out in an HN comment, GitLab shares had fallen nearly 50% from about 52ayearearliertoabout52 a year earlier to about 26 just before the announcement. Over the same period GitHub was absorbed into Microsoft and no longer trades separately, making direct comparison hard, but the fact that a Git forge SaaS — which should be the most natural beneficiary of the explosive growth in coding agent usage — has lost half its market value reads as a message from the market. To borrow simonw’s phrasing: “if investors’ fear is that AI makes GitLab’s business less valuable, then sliding this sentence into the GitLab Act 2 announcement is very rational.” The sentence he cites is this: “The agentic era multiplies software demand. (…) Last year, the developer platform market used to be measured in tens of dollars per user per month, this year it is hundreds/user/month and headed to thousands.”

If that sentence is true, GitLab has no reason to do layoffs. Tenfold or hundredfold revenue per user is grounds for hiring. So that sentence is closer to a signal to investors than a factual statement. HN’s torben-friis points to the same contradiction more bluntly: “To meet their largest opportunity ever, they believe they need less resources. I’m not sure I understand how that follows.” Staples himself seems aware of the contradiction. The answer he offers is that “We’re rewiring internal processes with AI agents, automating the reviews, approvals, and handoffs to speed us up.” To this torben-friis asks again: does this go in the bucket of “we make code twice as fast, the bottleneck is review, so YOLO and there’s no bottleneck”? “I have yet to see a convincing justification for that.”

A heavier comment lands on top. From dunder_cat: “CVE-2023-7028 (an account-takeover vulnerability where a semicolon between email addresses sent the password-reset email to two people) was exploited against my cluster. The bragging about fully automated changes and reviews scares me. The defective code wasn’t written by AI, but GitLab stands between many small organizations and their most precious resource. 2FA luckily saved us, but what will save us next time?” It is the most candid statement raising doubts about which direction the balance of the new “Speed with Quality” principle is tilting. That Speed precedes Quality in the word order could be a coincidence, or it could be intentional.

The cracks in the identity of a “handbook company” had been signaling earlier. The handbook page simonw cited still contains the sentence “Diversity, Inclusion & Belonging is one of our core values.” The reason he deliberately attached a permalink is that the sentence will soon disappear. A line being removed from the handbook is one thing; an entire value being lifted out wholesale is a first in the handbook’s 12-year history. More interesting is the old payroll.md page simonw also cites. GitLab once wrote out the salary structure for every country in its public handbook. A few years ago, that page was moved private. T being dropped from the values list did not happen overnight; it is the final step in a retreat that has been underway for several years.

shimman’s one-line comment is therefore harsh but accurate: “GitLab is a great example of a lifestyle company that should have never become a public corporation.” There is a sneer in the phrase “lifestyle company,” but inside it sits a structural diagnosis. A public company that has to answer growth questions every quarter on an earnings call cannot easily keep “we hold a transparent handbook as our identity” as a value all the way through. Handbook companies work best when they are private or non-profit. GitLab’s 12 years as a public company were a test of that proposition, and May 4 is the day the verdict was announced.

Another line of clues lies in the product itself. The comment thread is full of criticisms of GitLab’s product roadmap. usernametaken29 writes: “GitLab never stops surprising me with how bad their product roadmap is. Practical things like CI improvements are put off over UI rebranding on unicorn colours.” vultour is sharper: “For every issue I found, there was always a 4-to-7-year-old open ticket, and the only thing happening was that managers were randomly toggling labels on and off. With all the immense AI-driven developer productivity gains, surely we can finally close out these old issues quickly and clean the backlog. Or, alternatively, another AI-driven ‘workforce reduction.’ This charade is getting boring.” petetnt twists it once more: “Even with the current AI setup, GitLab hasn’t produced UX one could call great. I’m excited to see what they can make once the remaining human element is removed.” What these comments point at, collectively, is the diagnosis that GitLab’s real problem is not the quantity of staff but the direction of prioritization. And nowhere in the new operating principles is there an item called “reduce open issues.”

Section 3 — The Four-Cornered Map of Git Forge SaaS, and Beyond

Now let us place last week’s GitHub story and this week’s GitLab story on the same plane. In the last week of April, GitHub showed cracks across five axes at once — governance, uptime, security, cost, and supply-chain integrity. In the first week of May, GitLab announced layoffs and value abandonment together. The surface phenomena differ, but they point to one and the same question. Does the business model of hosting Git, layering CI, issues, and a package registry on top, and charging a monthly subscription per user — that is, Git forge SaaS — still work in the agentic era in the same shape?

Read again the sentence simonw quoted from GitLab’s post. The diagnosis is that monthly revenue per user is heading from tens to hundreds to thousands of dollars. If that diagnosis is right, it does not mean “per human user.” There is no way a single human consumes thousands of dollars worth of code hosting in a month. That revenue is, ultimately, the revenue that agents launched by humans pay on their behalf. In other words, the future customer of Git forge SaaS is not a human but an agent, and an agent creates many more PRs than a human, runs far more CI jobs, and is far less price-sensitive. If this hypothesis is correct, GitHub’s announcement of 30x load and GitLab’s “thousands per user” signal are two angles of the same landscape painted by two companies.

The problem is that nothing within that landscape requires the surviving shape to be SaaS. So fidotron’s comment is telling: “The fact they can’t capitalize on the current trainwreck of GitHub speaks volumes. If they had the right product people would be throwing money at them.” The vacancy this points to is not the vacancy of “a better GitLab” but the vacancy of “something that is neither GitHub nor GitLab.”

Let us re-examine the candidates for that vacancy. Codeberg is a non-profit, Forgejo-based forge, and has received meaningful migration inflows since the GitHub cracks. But Codeberg is, by definition, a place that rejects the “thousands per user per month” business model. What you get there is stability, not agent-era infrastructure. Self-hosted Forgejo·Gitea is operationally demanding, but in the agentic era some of that operational burden can be carried by agents themselves. That is a scenario in which self-hosting could recover some of its appeal. GitLab Self-Managed is the most interesting card. GitLab already has a Self-Managed license model, and changing the shape of the SaaS business does not eliminate the value of Self-Managed. If anything, while the handbook-style trust of SaaS GitLab is shaken, keeping code on your own infrastructure and merely buying a license could be reassessed as a partial safety line. And then GitHub — still the largest discovery pool, the largest agent-friendly infrastructure backed by Microsoft’s capital. Even with cracks, the kind of cracks that are hard to leave.

Nowhere in this four-cornered map is there a slot labeled “transparent handbook company.” That is the reason GitLab is vacating that slot and leaving.

bayindirh’s comment goes a step further: “Software will be built by machines under human direction. AI is the substrate on which future software will be built. Agents will plan, code, review, deploy, and repair.” In response to that line by Staples, he cites E. M. Forster’s 1909 short story “The Machine Stops”: “Honestly, I can’t believe how repetitively people ignore or are unaware of warning signs erected by people before them.” The comment is somewhat removed from infrastructure analysis, but the sentiment it gestures at — that a company’s identity can change so easily with a single announcement — runs through many other reactions in different vocabulary. If 12 years of handbook can be retired with one announcement, what was the weight of every promise written on top of that handbook?

TeeWEE’s comment hits the same point from another angle: “Git itself is distributed and was designed for scale. When they say ‘Git,’ I assume they mean ‘GitLab.’ A mistake that big couldn’t have slipped through. Are they planning to rebuild Git?” The post really does say: “Git itself is being reengineered for machine scale.” TeeWEE’s observation suggests two possibilities. First, if Staples really means to rebuild Git itself, the ambition is enormous. Second, if it is a typo for GitLab, that is sloppy IR copy. Either way, it is far from the precision of a handbook company.

Conclusion — The Audience of the Second Act

Return to the question at the start. Is the end of CREDIT a simple refresh of values, or a formal declaration of identity abandonment?

The answer is closer to the second. A company that refreshes values usually publishes a mapping of where each old value was absorbed into the new framework. GitLab’s announcement contains no such mapping. The fact that the post never says which corners of the new operating principles absorbed D and T is closer to saying they were not absorbed at all. The handbook’s D page will soon be edited; the word T will leave the values and survive only as an adjective. The two letters that made GitLab into GitLab for 12 years are vanishing in that way.

The strangeness of that vanishing lies in its happening publicly rather than secretly. A handbook company records even the retirement of its own identity in its handbook. simonw’s pre-attached permalink is therefore doubly meaningful. It preserves both a prior state of a single page and the paradox of the handbook as a genre. A company that has retired its values still attempts, through the transparency of the retirement process, to prove that it is still a transparent company.

Last week’s GitHub fractures and this week’s GitLab identity announcement are two faces of the same fact. For 18 years, Git forge SaaS has been the most stable infrastructure in a developer’s daily routine, and that stability was guaranteed in part by the values of the companies — one a giant SaaS under Microsoft, the other a transparent remote company built on top of a handbook. The agentic era is repricing those values. As humans cease to be the primary users of the infrastructure, values aimed at humans begin to look like costs.

The opening scene of the second act has ended. The company on stage still goes by the name GitLab, but the audience below the stage — those who for 18 years built their own companies’ remote policies by quoting that handbook — looks around the auditorium for the first time and asks the person next to them the same question. What was it, exactly, that we have been watching all this time?

Sources: