Undo

79% of engineering leaders say AI hasn’t made their releases any faster. An independent survey of 300 senior engineering leaders (most working on complex C/C++ codebases) explores why. → Where engineering time is really going now → Why coding agents struggle in complex codebases → How engineering teams reduce the risks of debugging complex issues with AI → How leaders weigh AI cost against capability → How your industry compares (semiconductor, EDA, networking, data management, finance) 📊 𝐑𝐞𝐚𝐝 𝐭𝐡𝐞 𝐫𝐞𝐬𝐞𝐚𝐫𝐜𝐡: Overcoming the Limitations of Coding Agents in Complex Software Systems.

Your engineers are writing code faster. So why aren’t they shipping faster? AI has transformed code generation. But in complex, mission-critical software, much of the hard work comes afterwards. Engineers still have to understand unfamiliar code, investigate failures, debug complex behavior and establish root cause. And that work is becoming a bigger part of the productivity equation. Our new research explores how 300 engineering leaders are experiencing the impact of AI on: → Code comprehension → Debugging and root cause analysis → Software quality → AI costs → Engineering productivity 📊 𝐑𝐞𝐚𝐝 𝐭𝐡𝐞 𝐫𝐞𝐬𝐞𝐚𝐫𝐜𝐡: Overcoming the limitations of coding agents in complex software systems.

Your engineers are writing code faster. So why aren’t they shipping faster? AI has transformed code generation. But in complex, mission-critical software, much of the hard work comes afterwards. Engineers still have to understand unfamiliar code, investigate failures, debug complex behavior and establish root cause. And that work is becoming a bigger part of the productivity equation. Our new research explores how 300 engineering leaders are experiencing the impact of AI on: → Code comprehension → Debugging and root cause analysis → Software quality → AI costs → Engineering productivity 📊 𝐑𝐞𝐚𝐝 𝐭𝐡𝐞 𝐫𝐞𝐬𝐞𝐚𝐫𝐜𝐡: Overcoming the limitations of coding agents in complex software systems.

Join us on 29th September to learn how engineering teams in your industry are fully automating root cause analysis. We'll cover zero-touch RCA in development, CI, and production (with no customer IP ever shared with a human).

Your AI agent didn't find the root cause. It found the most likely answer. That's a problem if you're working on mission-critical software. Whether you're developing networking software, silicon, EDA tools, or database products, quality isn't optional. AI can speed up development. But when diagnosing complex failures, today's agents still make educated guesses. On 5 August, Undo CTO Mark Williamson will explain why agents struggle with complex failures - and what leading engineering teams are doing differently to achieve reliable automated root cause analysis. Join us!

Why does AI still struggle to diagnose issues in complex software? Networking software, large-scale simulation models, EDA software, databases, trading systems... Engineers working on the world's most complex software report unreliable output, hallucinated APIs, poor systems understanding, subtle bugs, AI slop. Join this 30-minute session with Undo's CTO to better understand: 🪲 Why agents guess when debugging complex software ❓ Why larger models aren't the answer 🤖 How leading engineering teams are achieving fully automated root cause analysis