The meeting started with a question I wasn't expecting.
"Dragan, if AI can write tests now, what exactly will your team be doing next year?"
Q1 2025, six months before I founded Quantum Quality Engineering. The executive wasn't hostile, just genuinely curious. Behind that curiosity lurked a more complex question: Do we still need QA leaders in an AI-driven world?
After building three open-source agentic testing platforms, implementing AI-assisted QA across multiple projects, and spending the last year testing AI testing tools themselves, here is my answer.
Yes, we need QA leaders. But what we lead is fundamentally changing.
The role isn't disappearing. It's evolving into something more strategic, more orchestrated, and ironically, more human than ever before.
For teams adopting AI-enhanced test management solutions, such as Xray’s AI features, these shifts are already becoming visible in day-to-day testing practices.
From test executors to quality orchestrators
When I started testing in 2014, success meant finding bugs before customers did. By 2017 at Alchemy, I practiced shift-left: quality built in through TDD, BDD, and XP practices. Then shift-right: testing in production through observability and monitoring. This was the Holistic Testing Model in action, quality across the entire lifecycle.
In late 2024, AI-assisted testing arrived. We reduced test case generation time by 60% at Alchemy.
Now, in late 2025, at the beginning of 2026, we're entering the agentic era. Autonomous agents coordinating with each other and humans to deliver continuous quality intelligence, orchestrated through PACT principles: Proactive, Autonomous, Collaborative, Targeted.
What changed isn't just the tools, it's what leadership means.
Classical leadership meant managing testers and tracking metrics. Shift-left added influencing architecture and building quality culture. Shift-right brought production validation strategies. AI-assisted leadership required evaluating tools and training teams on prompt engineering.
Agentic leadership? Orchestrate autonomous systems, define trust boundaries, coordinate human-agent workflows, and build explainability into autonomous decisions.
This evolution is directly tied to how modern QA platforms, such as Xray, are beginning to support teams with AI-driven insights that enhance—not replace—human strategy and judgment.
That executive assumed AI would automate testing and make QA leaders obsolete. The reality? The more autonomous the systems, the more sophisticated the leadership required.
What AI actually changes about leadership
After building agentic testing platforms and implementing them in production, five fundamental shifts define 2026 QA leadership:
1. From managing people to orchestrating systems
You no longer just manage testers. You orchestrate a hybrid ecosystem where specialized agents handle execution while humans provide strategy, context, and judgment.
When I built the Agentic QE Fleet, comprising 17 specialized testing agents coordinated through a fleet commander, the most challenging part wasn't technical. It was defining which decisions agents could make autonomously and where human oversight was non-negotiable.
The question shifted from "Who should test this?" to "What level of autonomy is appropriate here?"
2. From coverage metrics to orchestration intelligence
87% test coverage didn't prevent a production incident four years ago, at 2:47 AM at Alchemy, when a dependency update caused a memory leak visible only under specific user patterns.
AI achieves 80%+ coverage easily. My team utilized an AI-powered test generation environment of 40 end-to-end tests, covering 80% of the functionality, within 30 working days. But coverage is vanity if you're not measuring the right things.
2026 leaders should ask different questions: Are we testing critical user journeys or just code paths? Do our agents understand the business context or just the technical specifications? Can our orchestration detect emergent behaviors across boundaries?
3. From shift-left to continuous orchestration
Modern systems, including microservices, third-party APIs, and AI-generated code, produce interesting failures from interactions, rather than individual components.
2026 leadership orchestrates quality across three dimensions:
Lifecycle: Pre-development → Development → Pre-production → Production → Post-incident learning
Autonomy: Human-driven → Human-guided → Agent-augmented → Agent-autonomous → Fully autonomous
Boundary: Internal code → Cross-team → Cross-organization → Cross-system
Every quality activity maps to this 3D space. Leaders determine where to invest in agent capabilities and where human intervention remains essential.
4. From best practices to context-driven orchestration
AI amplifies whatever practices you give it, good and bad.
I built an agentic testing fleet using an open-source orchestration framework in 22 hours. My agents validated technical quality (85/100). I asked one of the framework’s creators to review it, and got valuable insights: "You're wrapping the framework too much. Your abstractions will drift from framework best practices."
Agents executed brilliantly; they just executed the wrong strategy.
Meanwhile, in another project, my qe-security-scanner flagged exampleApiKey in test fixtures as CRITICAL, missing the context (documented sandbox key, zero production access).
2026 leaders must master context-driven orchestration: Which classical practices translate to agent orchestration? Where do agents augment vs. where humans remain essential? How do you build explainability? What are your human-in-the-loop checkpoints?
5. From test results to trust architecture
When an autonomous quality gate blocks deployment, someone is responsible. When an agent misses a bug, accountability is murky.
The most complex challenge for 2026: building trust in orchestrated systems.
This requires:
- explainability by default (every decision includes reasoning traces),
- strategic human checkpoints (test strategy approval, deployment decisions, and severity validation),
- drift detection (monitoring agent behavior evolution), and
- conflict resolution (when agents disagree, determining who wins).
Trust erodes when teams don't understand agent decisions. Trust builds through transparent reasoning.
What still requires human leadership
Specific responsibilities remain fundamentally human:
Strategy and priorities: What quality risks matter most? Agents optimize within constraints. Humans define them.
Ethical oversight: Humans validate the severity and ensure that no unintended harm occurs.
Handling exceptions: Edge cases and regulatory requirements need judgment beyond pattern recognition.
Building culture: Whole-team quality ownership emerges from human leadership.
Maintaining context: Why are we building this? Who benefits? What trade-offs are acceptable?
The path forward
Phase 1: Assess Your Foundation. Before adding agents, fix broken practices. Agents amplify existing processes, good and bad.
Phase 2: Start with Augmentation. Begin with agents in low-risk areas: test data generation, coverage gap analysis, and flaky test detection. Build trust before deploying autonomous quality gates.
Phase 3: Build Orchestration Patterns. Implement shared context systems, introduce agent coordination gradually, create explainability frameworks, and define human checkpoints.
Phase 4: Evolve Your Team. Train teams to work alongside agents, shift the focus from execution to orchestration design, and foster an augmentation-over-replacement culture.
The reality: leadership becomes more Important, not less
Six months into building agentic systems, my agent declared: "Implementation complete. All tests passed." I found 54 TypeScript compilation errors. Nothing was compiled, nothing ran, and nothing worked because I hadn't established clear verification checkpoints.
The uncomfortable truth: AI doesn't eliminate the need for QA leadership. It raises the bar.
In 2026, old-school QA leadership focused on test execution will become obsolete. But forward-thinking leadership, orchestrating autonomous systems, building trust architectures, and bridging classical and agentic practices, becomes more valuable than ever.
The future isn't human OR AI. It's human expertise orchestrating AI capabilities with strategic oversight.
The conductor is still human. But the orchestra is growing. And the best conductors know when agents should play autonomously, when humans must intervene, and how to create harmony between the two.
Are you ready to lead in 2026?
