AI: the great amplifier
Thoughtworks
Global tech consultancy blending design, engineering and AI to help businesses thrive through innovation.
Every organization is grappling with how to turn AI’s base metal into performing enhancing gold. That’s especially true when it comes to software development. We’ve seen some teams achieve incredible results but know that’s far from the universal experience.
The latest DORA report, of which Thoughtworks is a proud sponsor, adds illuminating detail on AI impact on software delivery. And the results bear pondering.
The extensive frontline research — from over 5,000 tech professionals across the globe — show how AI’s primary role is not as a replacement for developers but as an amplifier.
For the harmonious, high-performing teams, that’s great news. For those already under the pump, the picture’s less rosy: AI doesn’t solve broken systems — it can make them worse.
For leaders, the message is clear: investing in culture, alignment, and effective practices is more critical than ever. For teams, it’s a call to see AI not as a magic fix, but as a force multiplier for the habits and systems already in play. In short, AI amplifies what’s already there — the good, and the not-so-good.
How is AI impacting dev work?
Chris Westerhold explores the DORA report from a perspective of sustainable AI success, and how it requires systems thinking, integrated platforms, actionable insights and a strong engineering foundation. The value of engineers now lies in prompt engineering, solution architecture, and validating AI outputs.
📊 Read the full story here
Want to know more?
Don’t miss our deep dive on the latest DORA report when our own Chris Westerhold . Global Practice Director, Software Engineering Insights & Developer Experience, is joined by DORA report author and Google Lead Researcher Derek DeBellis .
Register here for a live discussion on the key findings of the 2025 DORA Report: State of Assisted Software Development.
Date & Time: October 21, 2025 — 4pm BST / 11am EST
Organizational design and Team Topologies after AI
How does AI impact organizational design? Our tech podcast team catches up with the Team Topologies authors to explore how AI is changing team capabilities, what it means for cognitive load and knowledge sharing and how to ensure there's structure and control without constraining experimentation and creativity.
🎧 Listen to the full episode here
- Stream it on Spotify here
- Stream it on Apple music here
Metrics that matter
Everyone knows: what gets measured gets managed. So if your tech teams are going to fly, you need to measure performance against the right metrics.
Check out this webinar replay to hear from experts from Thoughtworks , Google and DX on how to align metrics with roles, avoid common pitfalls and drive real outcomes. You’ll discover lively conversations about the incentives leaders select, the behaviors they reward and how that influences culture and performance.
🚀 Learn how to align metrics with outcomes — watch the replay now.
We hope you enjoyed this edition of Tech to know. Don't forget to subscribe for more.
Excellent analysis of the AI productivity paradox in the 2025 DORA Report. The insight about AI becoming a faster way to create chaos without proper foundations really resonates. I have seen this firsthand where velocity increases but comprehension decreases. The AI engineering waste categories you have identified are spot on. Prompt latency, context loss, toolchain fragmentation, validation overhead. These are exactly why dialogue engineering approaches are emerging instead of treating AI as a black box code generator. Your point about engineers evolving from code writers to solution architects and AI output validators is crucial. The real value shift is happening now. The organisations succeeding with AI are not just adopting new tools, they are fundamentally rethinking human-AI collaboration patterns. Thanks for highlighting how traditional productivity metrics break down with AI. The real question is not how much code did we generate but how much genuine capability did we build.
Jelle Taymans
Wondering how close Amplifier is to Parrot in LLM Vectorspace.
Thoughtworks make strong points, but the report already feels dated. Surely prompt engineering is dead, thanks to vastly improved interpretation. Two years ago, we needed 100 lines to prompt a digital scrum master. Last week, I got structured feedback from an AI impersonating an ISO 25010 cross-functional dev team… created using two lines of broken English after a couple of beers! Accelerating some, deprecating others? As an experienced engineer, AI lets me release high-value software faster than ever. But here’s the rub: many new devs aren’t engineers at all. They’re coders fluent in modern JavaScript or low-code platforms, with little understanding of what’s under the hood. We used to hand off designs and base classes for coding. Now? We can replace the whole lot with Cursor.AI. Which begs the question: where are new engineers coming from? These are the ideas I explore in my new blog, The Secret Sauce of Agile Development—following a live project with AI acceleration at the right moments. Check it out here: https://gc.charliebluster.com/articles/x1x1t4-the-secret-sauce-to-agile-develop With all the noise around AI, I honestly believe the core principles of Agile and engineering best practice are more vital than ever.