GitLab's Major Restructuring: Layoffs, Flattening, and a Bold Bet on the AI Agent Era

GitLab recently published a post titled "GitLab Act 2," announcing a series of major organizational and strategic changes. This isn't just an ordinary round of layoffs — it's a comprehensive strategic overhaul by a DevOps platform company bracing for the arrival of the Agentic AI era. Simon Willison offered an in-depth analysis, acknowledging GitLab's strategic vision while also casting a sober eye on its underlying motivations.

DevOps is a portmanteau of Development and Operations, representing a software engineering culture and practice that emphasizes collaboration between development and operations teams, along with continuous integration/continuous delivery (CI/CD). GitLab's core product is an all-in-one platform covering the entire software development lifecycle — from code hosting, code review, and CI/CD pipelines to security scanning and deployment monitoring — all integrated into a single application. Its business model primarily relies on a per-user seat subscription, with tiers including a free plan, Premium (approximately $29/user/month), and Ultimate (approximately $99/user/month). This means GitLab's revenue is directly tied to the number of paying developers — a crucial point for understanding the strategic narrative that follows.

Four Core Changes

Shrinking the Global Footprint: Cutting Operations in Up to 30% of Countries

GitLab has long been celebrated for its extreme remote-work culture, with employees spread across nearly 60 countries. Now the company plans to reduce the number of countries where small teams operate by up to 30%. This is a deeply significant decision — GitLab was once famous for its publicly transparent employee handbook, which even documented payroll processes for each country in detail (though this section was made private in 2023, the historical records remain accessible thanks to version control).

To appreciate the weight of this decision, you need to understand the deep roots of GitLab's remote-work culture. Since its founding in 2011, GitLab has operated as an all-remote company and is one of the largest all-remote organizations in the world. Its public GitLab Handbook spans over 2,000 pages, covering everything from onboarding procedures to asynchronous communication norms, and has been regarded as the "bible" of remote work. This radically transparent culture was one of GitLab's core competitive advantages, helping it achieve a successful IPO in 2021 at a valuation of approximately $15 billion. However, the compliance costs, time-zone coordination challenges, and cultural management burdens of cross-border remote operations have become increasingly apparent.

Shrinking the number of operating countries means lower management costs, but it also signals GitLab's shift from the idealism of "unrestricted remote work" toward a more pragmatic focus on operational efficiency.

Organizational Flattening: Removing Up to Three Layers of Management

GitLab announced it will remove up to three layers of management in certain functions, bringing leaders closer to the actual work. This trend isn't unique to GitLab — Coinbase previously announced an even more aggressive approach: no more than five levels below the CEO, with "no pure managers" — every leader must also be an excellent individual contributor, working like a "player-coach."

This trend traces back to earlier industry practices: Valve's flat structure, Spotify's Squad model, and Amazon's "two-pizza team" principle (teams should be small enough to be fed by two pizzas). Meta also conducted a massive management streamlining in 2023, with Zuckerberg calling it the "Year of Efficiency." The common logic behind these changes: as AI-assisted decision-making and automated coordination tools mature, middle management's traditional role as an "information router" is being replaced by technology.

The wave of management flattening is sweeping across the tech industry. The logic is clear: as AI agents take on more and more coordination and information-relay tasks, the traditional value of middle management is being reassessed.

R&D Restructuring: 60 Autonomous Small Teams

GitLab is restructuring its R&D department into approximately 60 smaller, more autonomous teams, each with end-to-end ownership — nearly doubling the number of independent teams. This philosophy aligns with what 37signals (parent company of Basecamp) has long advocated in its public handbook: "self-sufficient, independent teams" — each capable of delivering features independently without being blocked by other teams.

37signals, led by Jason Fried and David Heinemeier Hansson (DHH, creator of Ruby on Rails), has long championed a philosophy of small teams and low-complexity product development. The core of their Shape Up methodology is fully autonomous small teams (typically 2-3 people) working within 6-week development cycles, with complete decision-making authority from design to delivery. The advantage of this model is reduced cross-team dependencies and communication overhead, but the challenge lies in maintaining overall product consistency and architectural unity. GitLab's choice of 60 teams is clearly an attempt to balance autonomous flexibility with platform coherence.

Simon Willison noted that the evolution of AI agent engineering can indeed raise the capability ceiling of such small teams. When every developer can multiply their output with the help of AI agents, small and focused teams may actually hold an advantage over large, sprawling organizations.

Agentic AI refers to AI systems capable of autonomously planning, executing multi-step tasks, and adjusting behavior based on environmental feedback — distinct from traditional single-turn Q&A AI assistants. In software development, AI agents have evolved from simple code completion (like GitHub Copilot's early form) to autonomous systems capable of independently performing requirements analysis, code writing, test generation, bug fixing, and even code review. Notable examples include Devin (Cognition Labs), GitHub Copilot Workspace, and various coding agents built on large models like Claude and GPT. These agents are blurring the line between "developer tools" and "developer replacements" — which is precisely the core strategic tension GitLab faces.

Values Overhaul: From CREDIT to Speed and Ownership

Perhaps the most striking change is buried at the bottom of the announcement: GitLab is retiring its longstanding CREDIT values framework (Collaboration, Results for Customers, Efficiency, Diversity, Inclusion & Belonging, Iteration, Transparency), replacing it with three new values: Speed with Quality, Ownership Mindset, and Customer Outcomes.

The disappearance of "Diversity" from the core values is bound to draw attention. Interestingly, however, "embrace diversity, inclusion, and belonging" is still preserved as a sub-item under "Customer Outcomes." This adjustment looks more like a shift from declarative to actionable values — from "what we believe" to "what we do."

The Core Argument for the AI Agent Era: A Software Version of Jevons Paradox

In its announcement, GitLab put forward a highly compelling argument:

The agentic era will multiply the demand for software. Software has been the force multiplier behind nearly every business transformation of the last two decades. The constraint has been the cost and time to produce and manage software. That constraint is collapsing. As the cost of producing software collapses, demand for it will expand.

GitLab further noted that the per-user monthly fee for developer platforms is leaping from the tens of dollars to hundreds, and moving toward thousands. They believe there will be more software and more builders than ever before.

This is essentially Jevons Paradox applied to software: when the efficiency of using a resource improves dramatically, its total consumption increases rather than decreases. This paradox was first proposed by British economist William Stanley Jevons in 1865 in his book The Coal Question. He observed that James Watt's improvements to the steam engine dramatically increased the efficiency of coal use, yet Britain's total coal consumption rose rather than fell. The economic mechanism: efficiency gains lower the unit cost of use → more previously uneconomical use cases become viable → demand elasticity causes total usage to exceed efficiency savings.

This paradox has appeared repeatedly throughout technological history: LED bulbs are 80% more energy-efficient than incandescent bulbs, yet global electricity consumption for lighting continues to grow; cloud computing reduced server costs, yet total global data center energy consumption has surged. GitLab's argument is that AI agents will dramatically reduce the cost of software development, but the latent demand this unlocks will far exceed the cost savings — more businesses will be able to afford custom software, more business processes will be digitized, and total software output will explode.

The Interests Behind the Optimistic Narrative

However, Simon Willison also reminded readers to stay clear-eyed. GitLab's (NYSE: GTLB) stock price has fallen from approximately $52 to around $26 over the past year — a halving driven by market concerns about the outlook for its core business, as AI agents encroach on the developer tools market that GitLab depends on.

From a competitive landscape perspective, GitLab's position is particularly precarious. Its main competitor, GitHub (owned by Microsoft), has established a significant first-mover advantage in AI development tools with GitHub Copilot and Copilot Workspace, boasting over 150 million developer users. Additionally, emerging AI-native development tools like Cursor, Windsurf, and Replit Agent are redefining developer workflows from the ground up, bypassing the value chain of traditional DevOps platforms. The core threat GitLab faces is this: if AI agents can autonomously handle most development tasks, the traditional per-seat pricing model may face a fundamental challenge — when one AI agent can do the work of ten developers, will companies still need to buy ten seats?

If your entire business model depends on software engineering as a field continuing to grow and generating more high-value user seats, you have an extremely strong incentive to believe that AI agents will produce exactly that effect.

This is a classic case of motivated reasoning. GitLab's optimistic prediction about the AI agent era could be genuine strategic foresight based on deep insight, or it could be self-reassurance in the face of an existential threat. The truth likely lies somewhere in between. It's worth noting that GitLab's Jevons Paradox argument contains a key implicit assumption: that the price elasticity of software demand is large enough — meaning that cost reductions can unleash sufficient new demand to offset the decline in per-unit value. Whether this assumption holds will determine whether GitLab's strategic bet is visionary or wishful thinking.

Implications for the Industry

GitLab's restructuring reflects several common trends across the tech industry in the AI era:

1. Organizational structures are being redesigned for AI collaboration — smaller teams, fewer management layers, greater autonomy
2. The ideal of global remote work is facing a reality check — operational efficiency is beginning to take priority over geographic diversity
3. Corporate values are shifting from idealism to pragmatism — speed, ownership, and customer outcomes are replacing grander narratives
4. Platform companies face the dilemma of "self-disruption" — AI is both an opportunity and a threat, and the key question is whether the pace of transformation can outrun market changes

Regardless of how GitLab's bold bet ultimately plays out, this restructuring offers the entire industry a case study worth examining closely: when AI agents begin to reshape the fundamental paradigm of software development, how should a company built around developer tools reinvent itself?

Key Takeaways

• GitLab announced cuts to operations in up to 30% of countries, removal of up to three management layers, and restructuring of R&D into 60 autonomous small teams — fully embracing the AI agent era
• The company is retiring its CREDIT values framework; the new values focus on speed, ownership, and customer outcomes, with diversity demoted from a core value to a sub-item
• GitLab argues the AI agent era will multiply software demand through a Jevons Paradox effect, with per-user platform value leaping from tens of dollars to thousands
• GitLab's stock price has halved within a year; its optimistic AI agent narrative should be viewed critically in light of its own commercial interests
• The tech industry is seeing common trends of organizational flattening, small-team autonomy, and pragmatic values — AI is reshaping the very form of corporate organizations