24MAY

GitLab rewrites engineering around AI agents

4 min read

15:19UTC

GitLab CEO Bill Staples published 'GitLab Act 2' on 11 May 2026, cutting the company's country footprint by up to 30% and breaking R&D into approximately 60 smaller teams on the explicit premise that AI agents will plan, code, review, deploy, and repair software.

← Back to Rival studies split on AI's hit to jobs Jump to analysis ↓

EconomicDeveloping

Key takeaway

GitLab's Act 2 makes the CEO-signed case that AI agents replace engineering co-ordination, dissolving the management layer that connected them.

GitLab CEO Bill Staples published "GitLab Act 2" on Monday 11 May 2026, announcing a restructuring that cuts the company's country footprint by up to 30%, removes up to three management layers from some functions, and breaks R&D into approximately 60 smaller Teams, nearly doubling the number of independent Teams. The explicit justification in Staples's own words: "Software will be built by machines, directed by people. Agents will plan, code, review, deploy, and repair."

The 60-team structure concentrates individual-contributor output while stripping out the managers who previously co-ordinated across functions. In practice, individual engineers report into smaller pods with no manager between them and the AI tools they use to build product. A voluntary separation window closes 18 May; full restructuring completes by 1 June 2026. Financial impact lands on the 2 June earnings call, making it the first public headcount figure for the new structure.

GitLab did not arrive at this thesis alone. Salesforce CEO Marc Benioff applied the same structural argument to customer support: AI agents replaced roughly half the team, and no new engineers were hired in its fiscal year ending January 2026. Meta applied it to engineering titles in April, replacing conventional engineering roles with AI-native designations. GitLab has now applied the thesis to engineering structure itself: not which roles exist, but how many engineers it takes to ship product. That progression from support functions to engineering titles to engineering organisation is what makes the May cluster a pattern rather than separate decisions by unrelated companies.

The voluntary separation window that closes 18 May avoids the immediate WARN Act filing that a simultaneous termination event would trigger. GitLab operates across dozens of countries with no single US site large enough to meet the WARN threshold in any single jurisdiction. Both structural facts sit alongside the public manifesto, which is arguably the most candid CEO-signed statement on AI displacement to appear from a major DevOps company in 2026.

Deep Analysis

In plain English

GitLab makes software that other software companies use to build their products. Think of it as the tool companies use to organise engineers, review code, and push updates. On 11 May 2026, GitLab's chief executive published a document saying the company itself would restructure around the assumption that AI would do most of the engineering work going forward. His argument: AI agents can now plan what to build, write the code, check it, push it to production, and fix problems. Humans direct the process rather than perform it. So GitLab is cutting the number of countries it employs engineers in, removing manager layers, and splitting its engineering organisation into roughly 60 small, autonomous Teams. Why this matters beyond GitLab: GitLab is not a frontier AI lab predicting the future. It is a practical software tools company declaring that the future is already here in its own operations. When the company that makes the tools other engineers use restructures itself around AI-built software, it is saying something about where the whole software industry is heading.

Deep Analysis

Root Causes

GitLab's restructuring has two structural drivers that the fact does not name.

First, the GitLab product roadmap and the GitLab engineering structure have converged onto the same thesis. GitLab sells DevOps tooling including CI/CD pipelines, code review, and deployment orchestration. Its Act 2 document argues that AI agents will perform these functions.

If that thesis is correct for GitLab's customers, it must also be correct for GitLab's own engineering. The restructuring is therefore not a cost response to a revenue miss; it is the company applying its own product roadmap to itself.

Second, capital market pressure from the Harvard Business Review finding that only 2% of organisations announcing AI-related cuts have implemented AI that actually accounts for them creates a verification problem for companies that claim AI productivity gains.

GitLab, as a software tool company, has unusually direct access to internal metrics that can demonstrate AI productivity gains to investors. The 2 June earnings call is the first public test of whether those metrics materialise as disclosed.

What could happen next?

Precedent
GitLab's 60-team R&D structure becomes the reference architecture investors ask other mid-cap DevOps companies to justify not adopting.
Short term · 0.75
Consequence
The 2 June earnings call is the first public test of whether declared AI productivity gains translate to financial disclosure; a miss would undermine the entire CEO manifesto genre.
Immediate · 0.8
Risk
Removing co-ordination layers without a proven AI substitute creates integration risk that surfaces only after the voluntary separation window closes 18 May.
Short term · 0.7

Source Landscape

This story draws on neutral-leaning sources

Primary parallel: AT&T's 1996 restructuring into Lucent Technologies, NCR, and the rump AT&T broke a single 300,000-person engineering organisation into smaller, faster-moving units on the thesis that the internet era required a different operating model than the voice telephony era.

The structural logic was identical: concentrated individual-contributor output, fewer co-ordination layers, independent Teams. The results split: Lucent thrived in its first four years before losing $54 billion in the 2001-2002 correction; NCR returned to independence but never recovered its pre-split scale.

Counter-parallel: Amazon's own two-pizza team doctrine, which Upwork CEO Hayden Brown declared dead in the same fortnight, was also a structural response to a technology transition: Jeff Bezos introduced it in 2002 on the thesis that small Teams with full ownership produced better internet-era software than large integrated ones.

Amazon's team-sizing rule held for 22 years before an AI-era CEO publicly declared it obsolete. GitLab's 60-team architecture may be accelerating toward the same revision cycle, only faster.

GitLab has not disclosed the headcount number or financial cost of Act 2. The 30% country footprint reduction and three management layer removal are structural metrics, not salary figures. The earnings call on 2 June 2026 is the first point at which quarterly financial disclosure will produce a headcount number.

The economic read that is available before that disclosure: the 60-team structure, if it holds, means GitLab is restructuring its cost base around individual-contributor salaries (the Teams) and AI tool licences (the substrate) rather than management-layer salaries and co-ordination costs.

At $70,000-$120,000 average US engineering management compensation, removing three management layers from a company with 8,000 employees represents a structural labour cost reduction independent of any headcount cut. The voluntary separation window closing 18 May is the cleaner financial signal: how many senior engineers accept departure versus stay determines whether Act 2 is a genuine capability redesign or a managed departure.

Consensus view: Gartner Research Vice President Philip Walsh assessed in April 2026 that the 'AI-native pod' architecture GitLab describes reduces co-ordination overhead by removing manager layers, but only functions when the underlying AI toolchain is reliable enough to replace planning conversations. GitLab's own product is the toolchain in question, making it simultaneously the hypothesis and the test subject.

Counter-view: Boston Consulting Group's Henderson Institute argues that the two-to-three management layer removal targets precisely the function that keeps AI tool output coherent across Teams: the mid-level engineering manager who integrates AI output from multiple contributors, catches hallucinations, and manages scope. Removing that layer without a substitute mechanism transfers integration risk to individual contributors who have no organisational authority to resolve conflicts.

Key tension: Whether the 60-team R&D structure produces faster software delivery or fragmented output depends entirely on whether AI agents can replace the co-ordination function that the removed management layers previously performed, a capability no published benchmark yet measures at production scale.

Sources:GitLab (Bill Staples, CEO)

Mentions:GitLab →Bill Staples →Gartner →Boston →Marc Benioff →Meta →Amazon →Upwork →Hayden Brown →Salesforce →WARN Act →Harvard Business Review →Boston Consulting Group →

First Reported In

Update #9 · GitLab signs the manifesto, Brussels backs out

GitLab (Bill Staples, CEO)· 15 May 2026

Read original →

Causes and effects

Caused by

Salesforce halves support staff using AI

Salesforce's AI agent customer support replacement established the same structural thesis GitLab now applies to engineering itself

Occurred 29 Mar 2026

Read story →

Meta codes its own org chart

Meta's engineering title replacement with AI-native roles in April preceded GitLab applying the thesis to engineering structure itself

Occurred 19 Apr 2026

Read story →

This Event

GitLab rewrites engineering around AI agents

GitLab's restructuring applies the AI-replacement thesis directly to engineering structure, rather than headcount alone, making it the most operationally specific CEO manifesto in the 2026 tech restructuring cycle.

Different Perspectives

TSMC

The chipmaker raised its 2026 revenue growth forecast to about 44% on 15 October while holding capital spending at $60-64bn. High-performance computing, mostly AI data-centre chips, made up 68% of quarterly revenue, so demand for the hardware behind automation is still rising.

South Korean government

Its statistics office published the first youth figures on the 15-to-34 band set by a revised law, showing a fall of 98,000 against 147,000 on the old band. Including people in their early thirties, whose employment rose, makes the official youth decline look gentler.

Office for National Statistics

The ONS put UK vacancies at 691,000 for July to September and professional, scientific and technical vacancies 11.6% lower on the year. In September it had shown the same sector still adding jobs, so the fall appears in adverts before it reaches headcount.

Chinese government

Officials told a State Council Information Office briefing on 10 October that 10.52 million new urban jobs had been created by September, 87.7% of the annual target. Official figures show investment falling overall but rising in networks, data services and robot output, with urban unemployment at 5.2%.

Tharman Shanmugaratnam

Singapore's President, co-chair of the World Bank Jobs Council, urged governments at the Bangkok meetings to make crisis-era youth employment support a standing policy tool. That treats young workers' AI exposure as a lasting budget line rather than a shock to ride out.

Srini Pallia, Wipro

Wipro's chief executive called AI a long-term growth driver as the company cut 6,716 staff on flat revenue. HCLTech and TCS added staff in the same quarter, so Wipro's cut sits beside its own weak revenue rather than an industry-wide fall.

Key Figures

Country footprint reductionup to 30%

Management layers removedup to 3

R&D teams createdapproximately 60

In everyday terms

Software engineers who manage other engineers, or who co-ordinate between Teams, are the role category GitLab's restructuring specifically removes. The mid-level engineering manager role that co-ordinates between 6-10 engineers, manages scope, and integrates outputs from multiple contributors is the layer Act 2 targets. Engineers aged 28-40 in that range of responsibility, rather than the senior architect or the junior individual contributor, are the demographic where the immediate displacement is concentrated.

What’s Uncertain

Headcount and cost undisclosed until 2 JuneGitLab has disclosed structural metrics (country count, management layers, team count) but no headcount number or cost figure. The financial impact is unknown until the 2 June earnings call.

AI agent capability claim untested at production scaleGitLab's thesis that agents will plan, code, review, deploy, and repair has no published benchmark for production-scale reliability. The restructuring is premised on a capability that is asserted, not demonstrated.

Sources

1 centre

GitLab (Bill Staples, CEO)

Entities Involved

GitLab →Bill Staples →Gartner →Boston →Marc Benioff →

Evidence & sources

3 sources

SourceGitLab Act 2, Bill Staples, CEO, 11 May 2026
SourceHarvard Business Review, Thomas H. Davenport and Laks Srinivasan, 2026 {{EVREF:/t/ai-jobs-power-money/2/harvard-business-review-research-by-thomas-h-davenport/}}
SourceBoston Consulting Group Henderson Institute, 2026 (referenced in expert analysis)

Understand the Context