Justin Reock is Deputy CTO at DX, where he studies whether AI coding tools actually improve software delivery, code quality, and business performance. His work challenges engineering leaders to distinguish genuine organizational gains from enthusiastic adoption, misleading productivity metrics, and expensive increases in generated code.
From open-source architecture to engineering productivity
Reock built his career around the systems that shape developers’ work. As chief architect at Perforce and a technical leader at OpenLogic, he helped enterprises navigate Java, messaging infrastructure, observability, software licensing, and open-source adoption. His contribution to the 2019 Open Source Support Report connected enterprise open source with machine learning and cloud-native infrastructure.
By 2023, he was Field CTO and Chief Evangelist at Gradle, focused on developer productivity engineering: build performance, test infrastructure, and feedback loops. His work on build-system migrations showed how observability can expose regressions and reduce the disruption of major toolchain upgrades. He subsequently became Head of Developer Relations at Cortex, working on internal developer portals, engineering scorecards, and organizational ownership.
At DX, Reock applies that experience to the entire software-delivery system, combining engineering telemetry with developers’ firsthand experience to evaluate AI’s practical effects.
- Measure utilization, impact, and cost. Reock’s AI measurement framework separates adoption from changes in delivery, quality, and spending. He combines product telemetry, developer surveys, and targeted experience sampling, tracking indicators including pull-request throughput, change failure rates, maintainability, and confidence in production changes.
- Find the actual bottleneck. Drawing on the Theory of Constraints, Reock argues that faster code generation accomplishes little when work stalls in planning, review, onboarding, incident response, or coordination. His analysis of engineering productivity data associates growing AI usage with substantially smaller gains in pull-request throughput.
- Account for downstream costs. His 2026 assessment of AI’s engineering impact highlights larger pull requests, rising spending, uneven productivity improvements, and the risk that increased output creates additional review burdens or technical debt.
- Build reliable human-agent workflows. Reock’s advanced prompting guidance emphasizes structured prompting for complex systems, validation loops, independent implementations for critical work, and tightly scoped refactoring. He advocates maintaining shared agent instructions, protecting experimentation time, involving compliance teams early, and treating psychological safety as essential to effective adoption.
