More AI. More code. More problems?
Why the real bottleneck isn’t where you think
May’s Monday Merge is here, and this month, we’re sharing a question many teams are beginning to ask:
What happens when AI generates more code… but the rest of the software development lifecycle can’t keep up?
Because more code doesn’t always mean more progress.
In fact, for many teams, it’s revealing something new… a hidden tax.
More code to review. More vulnerabilities to fix. More pipelines to manage. And not enough context to keep everything moving.
That’s exactly the problem GitLab is focused on solving.
Because the real opportunity isn’t writing more code faster… it’s fixing what slows it down.
That’s where GitLab 18.11 comes in 👇
GitLab 18.11 — Tackling the AI Paradox
With GitLab 18.11, we’re continuing to push agentic AI beyond code generation… and into the parts of the development lifecycle where work actually gets stuck.
At the heart of this release is Agentic SAST Vulnerability Resolution, now generally available on GitLab Duo Agent Platform.
Instead of just flagging issues, it:
- Analyzes vulnerabilities in context
- Generates a fix
- Opens a ready-to-merge request with a confidence score
All before those vulnerabilities reach production.
We’ve also introduced two powerful new agents designed to remove friction from everyday workflows:
- Data Analyst Foundational Agent (GA) → Ask questions in natural language and get instant, visual insights into your software lifecycle data, from pipeline performance to delivery trends
- CI Expert Agent (Beta) → Scans your repo, understands your setup, and proposes a ready-to-run CI pipeline in minutes — no more starting from a blank YAML file
And as teams scale AI across the business, we’re also introducing new controls for GitLab Credits, giving organizations clearer visibility and guardrails around AI usage and spend.
Because solving the AI paradox isn’t just about accelerating code generation…
It’s about reducing bottlenecks, accelerating decisions, and keeping the entire software lifecycle moving.
Recognized for the full lifecycle, not just code
This shift isn’t going unnoticed.
GitLab has been named a Leader in the 2026 Omdia Universe for AI-assisted software development.
And the interesting part here is why.
For the first time, vendors aren’t just being evaluated on coding capabilities… They’re being assessed on their ability to support the entire software lifecycle.
From planning and testing to security and deployment.
And that’s where GitLab stands out.
Because real acceleration doesn’t come from isolated tools. It comes from orchestration.
☁️ GitLab + Google Cloud, bringing context to AI
Another big step forward is our collaboration with Google Cloud.
By integrating GitLab Duo Agent Platform with Vertex AI, teams can bring powerful models into a fully governed DevSecOps workflow.
What does that actually mean?
It means:
- AI agents operate with full lifecycle context
- Teams stay within a single system of record
- Governance, security, and model control are built in
Instead of AI working in silos, it works across your entire SDLC, from planning to remediation to delivery
And that’s where things start to scale in a meaningful way.
Customer spotlight: Rivian & the future of software-defined vehicles
To see what this looks like in the real world, GitLab CEO Bill Staples recently sat down with Rivian’s Eric Hulser.
The scale they’re operating at is incredible.
We’re talking:
- Thousands of jobs running every second
- Massive validation pipelines
- Software that directly impacts real-world safety
Recommended by LinkedIn
As Eric puts it,
every piece of code that goes into a vehicle must be highly tested, reliable, and safe.
This is a powerful reminder that in some industries… software quality isn’t just important. It’s critical.
Here's the full interview...
Transcend is back!
June 10 — London (The Landmark, Marylebone) + Virtual livestream
Transcend returns as a one-day experience uniting leaders, researchers, and developers to tackle one of the biggest questions in our industry right now:
How do you scale AI without creating more complexity?
This year’s event is all about closing the gap between AI speed and real-world delivery.
You can look forward to:
- Brand new research into the growing challenge of AI
- Live demos of agentic workflows across the software lifecycle
From executive keynotes and customer stories to hands-on workshops and live developer sessions, Transcend is designed to be more than just a conference… it’s a glimpse into what’s coming next.
If you’re serious about where AI is heading in software development, this is one you won’t want to miss.
🗓️ Mark your calendars and register here!
On the road with AWS
We’re out hitting the road throughout May, teaming up with AWS to bring DevSecOps to life across multiple events.
We’re kicking things off at AWS Summit Stockholm on May 7 with a full day of hands-on sessions, live demos, and conversations around the cloud and AI technologies shaping the future of software development.
From there, we’re heading across the US for our SLED DevSecOps Roadshow, with stops in Sacramento, New York, and Chicago.
Across both experiences, you can expect:
- Hands-on demos of GitLab in action
- Real-world use cases from teams scaling securely
- Opportunities to connect with peers solving similar challenges
Whether you’re exploring AI-powered development or looking to simplify and secure your software delivery, these events are all about showing what’s possible when everything works together.
What we’re reading
We’ve had our heads down in some seriously insightful reads this month…
🔍 The hidden tax of AI-generated merge requests Brian Wald explores how AI is shifting bottlenecks from writing code… to reviewing it.
As merge request volume increases, review capacity doesn’t scale the same way which puts pressure on senior engineers and slows down delivery.
Jamie Dicken shared why security’s operational maturity can determine whether a new tool becomes a productivity driver or expensive shelfware.
♟️ A thought to leave you with
We really have a lot happening right now. Agentic AI is opening up huge new possibilities… but also forcing us to rethink how we build, secure, and deliver software.
And it got me thinking about this:
“Just as the steam engine extended human physical capabilities, AI can extend our mental capabilities.” ~ Garry Kasparov
Kasparov isn’t just any commentator on AI. He’s a chess Gransmaster and former world chess champion who famously faced off against IBM’s Deep Blue in one of the earliest and most symbolic battles between human intelligence and machines.
Since then, he’s become a strong advocate for human + AI collaboration… not replacement.
A reminder that AI isn’t here to take over, it’s here to amplify what we’re capable of... but only if we build the systems around it the right way.
Thanks for reading. The future isn’t just faster code generation, it’s everything working together.
See you in the flow and, as always, happy merging!
Fatima Sarah Khalid | Sr. Developer Advocate, GitLab.
Like this newsletter? Forward it to your team. And don’t forget to 👉 subscribe to our YouTube channel!
The hidden tax of AI-generated code isn't just about post-coding tasks; it's also about maintaining a cohesive development lifecycle. GitLab's agentic capabilities are key, but we mustn't overlook the importance of context and real-time feedback in preventing bottlenecks from forming in the first place.
This is the part most teams underestimate. AI increases output, but it also increases the load on review, security, and pipeline coordination. Without context, that work doesn’t scale with the code. That’s where things start to slow down.
Definitely worth reading.