Abstract
AI, productivity, and the parts of software that didn’t get easier
Generative AI can help us produce code faster than ever before, but faster code generation does not automatically translate into faster or safer delivery. In practice, AI acts as an amplifier: teams with strong engineering fundamentals improve, while teams with existing bottlenecks and weak software development practices feel those problems more acutely.
In this talk, we’ll look past the hype and focus on what actually drives productivity in an AI-accelerated world. We’ll explore why shiny new tools don’t fix broken systems, why optimising individual steps rarely improves end-to-end flow, and why software engineering fundamentals matter more — not less — when code is easy to generate.
This is not an anti-AI talk. It’s a reminder that the hard parts of software didn’t get easier, and that fundamentals determine whether AI makes us faster, or breaks us faster.
Videos
I'll post the video when it's available
Books
Contains affiliate links - the price to you is not affected, but I may receive a small commission if you by via the link.
Core Engineering Practices & Development
- Modern Software Engineering
- Extreme Programming Explained
- Tidy First?
- Agile Testing
- Test-Driven Development: By Example
- Growing Object-Oriented Software, Guided by Tests
- User Stories Applied
- Working Effectively with Legacy Code
- Vibe Coding
- 97 Things Every Java Programmer Should Know - yeah I know I edited this book. You should still read it if you do Java, it has some intentionally conflicting advice because writing code is messy and you need to be able to hold several points of view at the same time
Delivery & Production
- Continuous Delivery
- Accelerate
- Release It!
- Observability Engineering
- The DevOps Handbook
- The Phoenix Project
Design, Architecture & Maintainability
- The Mythical Man-Month
- Domain-Driven Design
- Design Patterns
- Head First Design Patterns
- Building Evolutionary Architectures
- 97 Things Every Architect Should Know
- The Software Craftsman
Organisations, People & Productivity
- The Lean Startup
- Peopleware
- The Pragmatic Programmer
- Slow Productivity
- Drive
- Organizational Physics
- Ergodicity
Links
Here's a mish-mash of things I read, watched or listened to when I was preparing this talk. Many of them have themes that didn't make it into the talk, but still contributed to my thinking on the topic. Chat GPT grouped them for me so your mileage may vary,
1. AI makes coding faster (but that’s not the win)
- The AI Productivity Paradox in Software Development—Why Developers Feel Faster But Measure Slower - "Think about Amdahl’s Law: system performance is limited by the slowest component. Even if code generation accelerates dramatically, the system cannot move faster than its bottlenecks—review, testing, deployment."
- The AI Productivity Paradox Report 2025
- The AI Layoff Trap (Research Paper) - "AI will destroy the economy"
- The developer productivity paradox: Why faster coding doesn’t mean faster software delivery - oh yeah I wrote this.
Core idea:
AI improves local efficiency (writing code), but not system throughput.
2. The real bottlenecks were never coding
- Writing Code Was Never the Bottleneck
- Harness engineering for coding agent users
- Patterns for Reducing Friction in AI-Assisted Development
- It Doesn’t Help To Push AI Into A Crappy Process - Emily talks about how what we were doing before is "Code Driven Development" and we need to be doing software engineering
Core idea:
Delivery speed is constrained by systems: feedback loops, environments, testing, integration.
3. AI exposes (and amplifies) system constraints
- Zendesk Says AI Makes Code Abundant, Shifting the Bottleneck to “Absorption Capacity”
- The LeadDev AI Impact Report 2025 - "Regardless of their actual effectiveness, developers like using these tools, and they are here to stay. According to a recent McKinsey survey, engineers find that, with generative AI, they are happier, more able to focus on satisfying and meaningful work, and more able to achieve “flow state.”"
- State of AI in Business 2025 Report - Despite $30–40 billion in enterprise investment into GenAI, this report uncovers a surprising result in that 95% of organizations are getting zero return."
Core idea:
Teams can only absorb so much change — AI increases input, not capacity.
4. The role of the developer is shifting (not disappearing)
- When AI writes almost all code, what happens to software engineering?
- Why AI Makes Software Engineering Harder, Not Easier - I really liked this article, and when I first read it I had the impression that Markus was saying basically everything I wanted to say in this talk
Core idea:
Value shifts from typing code → understanding problems, intent, and systems.
5. Humans are still the critical factor
- Is It Still All About the People?
- What 30,000 YouTube Comments Tell Us About AI and Software Engineering - "This tells us something worth paying attention to. What we might call "AI anxiety" is not really a standalone phenomenon. It is part of a wider professional anxiety conversation that was already running"
Core idea:
Collaboration, communication, and judgment still dominate outcomes.
7. Cultural narrative: hype vs reality
- The 3 Words That Secretly Drive Developer Productivity & Motivation - (also read Daniel Pink's Drive)
- How AI will change software engineering – with Martin Fowler
Core idea:
The industry narrative is dominated by transformation stories, not system thinking.