From faster coding to accelerated innovation cycles: How intelligent orchestration unlocks AI's promise

Something fundamental changed in late 2025 that altered the trajectory of the enterprise AI market. 

Three AI model releases — Gemini 3 by Google, Opus 4.5 by Anthropic, and GPT-5.2 by OpenAI — crossed a capability threshold that has industry leaders rethinking the role of AI in coding. Gergely Orosz reports in his Pragmatic Engineer newsletter that experienced engineers are now letting AI generate 90%+ of their production code.

However, that fundamental change is not just about AI writing more code. It’s about how the hundreds of steps in the business-critical process of software development — ensuring quality, security, compliance, and maintainability — can be sped up with more AI-generated code. 

With AI, organizations report a 48% increase in developer productivity, with developers generating functions, debugging errors, and implementing features at unprecedented speed. But coding represents only 20% of the software delivery cycle. According to Amdahl's Law, making coding, say, 10x faster, yields only a 1.25x overall speedup when the remaining 80% of the cycle, including code review, testing, security, and deployment, remains unchanged.

Organizations that adopt AI only for coding will hit a wall. Their teams will report productivity gains while drowning in review backlogs, flaky tests, compliance requirements, and performance optimization.

GitLab calls this the 'AI Paradox.' While AI makes coding faster, the greatest opportunity to accelerate innovation lies in enhancing quality, security, and speed throughout the entire software lifecycle.

From sequential stages to continuous loops

Traditional software delivery operates in discrete stages with manual handoffs: plan, code, test, secure, deploy, operate. Each handoff introduces delays, context loss, and coordination overhead. When AI accelerates only one stage (coding), the handoffs become the bottleneck. Each transition between stages destroys the velocity gains from faster coding.

Intelligent orchestration replaces this sequential model with continuous execution loops. Instead of "code, then test, then secure," teams continuously generate, test, secure, deploy, and verify work in parallel. Agents execute autonomously within this continuous flow while humans orchestrate from above, setting direction and governance without getting pulled into every tactical decision.

This shift eliminates the gaps between stages that slow team velocity. Work no longer waits in queues before the next stage begins. Context persists throughout the loop rather than being lost at each handoff.

Solving the AI Paradox requires intelligent orchestration built on three foundational pillars:

Workflows: Teams and AI Agents Working Together. Software teams define the rules for AI agents, including which context to rely on, workflows to streamline, and compliance rules to enforce. They will move beyond one-to-one AI chat experiences to team-level agentic workflows where multiple agents collaborate on complex tasks, issue-to-merge-request flows, security analysis, code reviews, and CI/CD operations. In this scenario, one agent serves numerous developers, multiple agents work in parallel across teams,  and humans steer and guide rather than micromanage individual AI outputs.

Context: Unified Data and Intelligence Across the Lifecycle. Instead of sequential handoffs that break flow, intelligent orchestration maintains continuous execution across stages through a unified data model. Unlike fragmented tools where context is lost across systems, this provides complete context across the entire lifecycle. Teams gain visibility into not just code, but requirements, history, security implications, deployment constraints, and operational feedback. As a result, teams work on multiple projects and releases simultaneously without losing context. No stage waits for another;  what used to be a painful slog through tickets, waiting, handoffs, and remediation sprints becomes continuous generation, continuous compliance, and continuous improvement.

Guardrails: Governance and Compliance Built Into Flow. Flexible deployment options with custom rules for security and compliance provide full control over your data and workflows. Agents assess risk and recommend appropriate levels of autonomy for each task, with policy-driven guardrails that ensure higher-risk changes receive more human oversight, all in a single orchestrated system. This enables teams to maintain velocity without sacrificing security or compliance, which are built into the flow and automatically enforced rather than bolted on afterward.

The human element

The solution isn't more tools or faster AI. It's rethinking how humans and AI work together and fundamentally redesigning the software delivery process itself. GitLab's research shows that 76% of DevSecOps professionals believe AI will create more engineers, not fewer. What’s changing is the nature of the work itself.

When AI writes most of the code, the skills that become critical are those that used to be expected of senior or staff-level engineers: breaking work down into well-defined pieces, making sound architectural decisions, taking a product-minded approach, mastering automated testing and observability, and tracking tech debt. As Gergely Orosz observed, "Tech lead traits will almost certainly be more in demand. When AI can implement any well-defined ticket, who will write the ticket that makes AI correctly create the code?"

In practice, AI agents handle repetitive tasks autonomously across multiple stages, generating code, running tests, scanning for vulnerabilities, and deploying changes. Humans set direction, maintain governance, and make judgment calls. This shift moves developers from writing every line of code to orchestrating systems and guiding AI agents. Valuable human skills such as creativity, strategic vision, judgment, and architectural thinking, become more critical, not less.

At Ericsson, a leading telecommunications company and GitLab customer, managing enterprise software deployments across more than 300 global communications service providers requires seamless orchestration across multiple tools, systems, and workflows. After adopting a unified platform approach, Ericsson achieved 50% faster deployments and saved 130,000 hours over six months, enabling it to deliver updates in weeks rather than months.

This pattern repeats across industries. Indeed achieved a 79% increase in daily software development pipelines while lowering its hardware costs by up to 20%. At CERN, 10,000 scientists from over 100 countries collaborate on particle physics research, achieving 90x faster job startups. Lockheed Martin retired thousands of Jenkins servers and accelerated software delivery, moving from monthly to weekly deployments with security and compliance built into workflows.

These aren't stories about AI coding tools making individual developers faster. These are stories about intelligent orchestration enabling teams to maintain velocity at enterprise scale.

From “AI Paradox” to opportunity

AI has accelerated software development, but so far only at the individual coding level. Intelligent orchestration extends that acceleration across entire teams and the complete software lifecycle.

Enterprise leaders evaluating their software delivery approach need to determine whether their tools enable teams to accelerate innovation or slow it down. When software teams and AI agents work in flow together, they become vision-driven rather than logistics-driven, shifting their focus from "Can we coordinate this?" to "What should we build next?"

Manav Khurana is chief product & marketing officer for GitLab.