AI can help developers produce code quickly without making the software-delivery process faster. GitLab’s 2025 and 2026 surveys describe an “AI Paradox”: code generation accelerates while review, testing, security, compliance, deployment and handoffs remain bottlenecks—or become more pressured as more code enters the pipeline.
What GitLab means by the “AI Paradox”
GitLab uses the term to describe a gap between local coding speed and end-to-end delivery speed. In a November 10, 2025 release, GitLab chief product and marketing officer Manav Khurana said: “This survey illustrates what we call the ‘AI Paradox,’ where coding is faster than ever, yet the lack of quality, security, and speed across the software lifecycle is causing friction on the road to innovation,” GitLab’s release says.
The distinction matters because shipping software includes much more than writing source code. GitLab’s March 2026 explanation characterizes coding as about 15% of shipping work, with code review, testing, security scanning, compliance and deployment making up the other 85%. That is GitLab’s framing, not an independent time-and-motion study, but it illustrates why a faster coding step may not shorten the full path to production.
What the 2025 survey reported
The 2025 Global DevSecOps findings came from The Harris Poll for GitLab and covered 3,266 DevSecOps professionals in IT operations, IT security and software development. GitLab’s published release does not provide the field dates, sampling frame, response rate, weighting or margin of error, so these figures should be read as reported survey responses rather than universal measurements.
#1 Best Overall
| Finding | GitLab’s reported result | What it indicates |
|---|---|---|
| Process and collaboration waste | 7 hours per team member per week | Respondents said inefficient processes and collaboration barriers consume this time; it is not a measured average for every team. |
| Development-tool fragmentation | 60% use more than five software-development tools | More systems can mean more context switching, integrations and handoffs. |
| AI-tool fragmentation | 49% use more than five AI tools | Multiple assistants and agents can create inconsistent controls and records. |
| Production cadence | 82% deploy at least weekly | Frequent releases increase the importance of automated validation and evidence. |
| Compliance pressure | 70% agree AI makes compliance management more challenging | Teams perceive additional governance work as AI use expands. |
| Late discovery | 76% say more compliance issues are found after deployment than during development | Controls may be arriving too late to prevent rework. |
| AI adoption | 97% use or plan to use AI in the software-development lifecycle | AI is becoming broadly relevant to delivery policies. |
| Need for human review | 37% would trust AI to handle daily work tasks without human review | Most respondents did not express that level of unconditional trust. |
| “Vibe coding” problems | 73% report problems with code created through the term’s “vibe coding” approach | Fast generation does not guarantee maintainability, security or correctness. |
GitLab also wrote that “Toolchain fragmentation has created bottlenecks for developers, and AI agents are amplifying the issue.” The survey supports a concern about friction and perceived risk; it does not prove that AI itself causes every team’s slowdown.
Why more generated code can slow delivery
Review queues grow
Generated code still needs someone to establish intent, check edge cases, assess maintainability and approve the change. If production capacity rises faster than reviewer capacity, pull requests wait longer and reviewers face larger batches.
Testing and security become the constraint
Every additional change needs automated tests, dependency checks, static analysis and secret detection. A team that generates code faster without expanding validation capacity can move defects and vulnerabilities downstream instead of shortening lead time.
Compliance evidence arrives late
Regulated delivery often requires traceable approvals, policy checks and records of what changed. GitLab’s 2025 respondents reported compliance challenges and more issues discovered after deployment, a pattern that can force expensive rework or release delays.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchHandoffs erase local gains
Developers may finish a change quickly, but security, quality, operations and legal or compliance teams can remain sequential gates. Fragmented tools make status, ownership and evidence harder to find, increasing coordination time.
What the separate 2026 survey adds
GitLab released a different survey on June 23, 2026. The Harris Poll surveyed 1,528 developers and technology buyers across six countries. These respondents, questions and denominators are not the same as the 2025 study.
| 2026 finding | Share |
|---|---|
| Developers write and commit code faster after adopting AI tools | 78% |
| Individual productivity improved while overall delivery did not accelerate at the same pace | 79% |
| AI shifted the bottleneck from writing code to reviewing and validating it | 85% |
| Reported some governance challenge with AI-generated code | 92% |
| Adopted AI tools faster than governance policies were developed | 80% |
| Could not reliably distinguish AI-generated from human-written code in their codebase | 43% |
The governance findings sharpen the provenance problem: organizations may need to know how code was produced, which policies applied, what approvals occurred and whether a human can still account for the result. They remain vendor-reported survey results, not proof of a universal causal relationship.
How to test whether AI improves your delivery
Measure the complete workflow against a baseline before adding more assistants or agents. Track the same services and release types for a defined period, and segment results by team or system rather than relying on a single company-wide average.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
- Delivery lead time: time from a first code change to production.
- Review queue time: how long a change waits before review and approval.
- Validation time: duration and failure rate for tests, security scans and policy checks.
- Deployment frequency: successful production deployments over a period.
- Change failure and escaped defects: rollbacks, incidents and bugs discovered after release.
- Handoff delay: waiting time between development, security, operations and compliance.
- AI traceability: whether the team can identify generated code, model or agent involvement, approvals and applicable policy.
A faster commit rate is a useful leading indicator only if these end-to-end measures improve without unacceptable increases in defects, incidents or review burden.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Practical responses to the bottleneck
DevOps modernization
Audit where work waits and which tools exchange information poorly. Consolidate source control or CI/CD only where fragmentation is materially slowing delivery, and standardize reusable pipeline patterns. Consolidation is an option, not a guaranteed remedy; judge it by lead time, failure rate and handoff data.
Security modernization
Run dependency scanning, static analysis and secret detection in the delivery pipeline. Move policy enforcement and evidence collection earlier and make them continuous, so a release does not become the first point at which a problem is found.
AI modernization
Expand beyond individual code suggestions only after workflow, security and governance controls are ready. Define which changes require human approval, retain provenance records for agent activity and establish a rollback path. AI should increase validated throughput, not merely the volume of unreviewed code.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →GitLab presents these three paths as connected modernization choices. Its customer examples, including a claim that Ericsson saved 130,000 engineering hours in six months, are vendor-reported case claims and should not be treated as expected results for another organization.
A decision framework before buying or expanding AI tools
| Question | Evidence to require |
|---|---|
| Which constraint is being addressed? | A measured review, testing, security, compliance or handoff bottleneck—not a general promise of productivity. |
| What is the integration cost? | Pipeline changes, identity and access work, data movement, training and ongoing administration. |
| Does it support policy and audit needs? | Approval records, configurable rules, evidence retention and clear ownership. |
| Will review capacity improve? | Automation that reduces repetitive validation without removing accountable human decisions. |
| Can generated code be traced? | Useful records of AI or agent involvement, prompts or actions where appropriate, model version and approvals. |
| What end-to-end outcome will change? | A baseline and target for lead time, deployment frequency, defects, incidents and waiting time. |
The bottom line for software teams
GitLab’s surveys show why “AI makes coding faster” is an incomplete productivity claim. In the 2025 study, respondents reported fragmented tools, compliance difficulty and limited willingness to remove human review. The separate 2026 study reported faster coding alongside slower overall delivery, a review-and-validation bottleneck and governance gaps. Treat those results as signals to inspect your own workflow, not as proof that AI inevitably slows every team.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
