The Productivity Paradox: What AI Does to Engineering Output
Developers can feel 20% faster with AI and still be 19% slower. This session explains why faster coding doesn’t automatically mean faster delivery – and how to measure AI value through shipped, stable software instead of activity.
Speakers
Roman Gorge
Head of Cloud Partnership at Andersen
In his role at Andersen, Roman combines his leadership skills, international experience, and deep knowledge of the cloud, AI, and data to solve the business and technical challenges of our clients.
Igor Volkov
Head of Engineering Practices and Architecture Practice Head at Andersen
Igor has 20+ years in enterprise software across fintech, telco, and healthcare. Former Principal Solutions Architect at Netcracker, delivering BSS/OSS transformations for Tier 1 global telcos.
Marat Kornyev
AI PM, Lead AI Consultant at Andersen
Marat is an AI Specialist and Project Manager specializing in AI, automation, and digital product development. Passionate about turning emerging technologies into practical, scalable solutions.
Agenda
Defining AI Value Before You Claim It
- Before You Celebrate the Win: How to Actually Measure AI's Payoff. Most teams claim an AI productivity win without ever defining what "value" means.
- You Feel 20% Faster. The Data Says You're Not. A closer look at the disconnect between developer sentiment and measured throughput: why autocomplete removes friction, not effort, and why that gap quietly inflates every AI ROI report.
- Lines of Code Lie: What Should Actually Count as Progress. Velocity and code volume are easy to track and easy to game. The case for measuring what reaches production - stability and shipped outcomes - instead of activity at the keyboard.
Roman Gorge
Andersen
Why Faster Coders Don't Mean Faster Delivery
- One Faster Machine Doesn't Speed Up the Whole Factory. Coding is one station in a much longer delivery line: reviews, coordination, and handoffs are the rest.
- Typing Was Never the Bottleneck. If writing code is a minority of an engineer's week, speeding it up can't fix delivery on its own.
- AI Didn't Remove the Bottleneck. It Just Moved It. As AI accelerates code generation, review queues and coordination overhead absorb the pressure instead. How that shift stresses agile workflows and what teams can do to keep pace without losing control.
Igor Volkov
Andersen
Live Q&A
Roman Gorge
Andersen
Igor Volkov
Andersen