AI is the headline — and rightfully so. But speed in software rarely arrives from one magic feature. The more useful story, underscored by Nicole Forsgren and Abi Noda in their book Frictionless and explored in an excerpt on Pragmatic Engineer, is that the fastest, most reliable way to get business outcomes is to remove developer friction.
Why friction matters — and why AI alone won’t save you
Every executive loves the promise of tools that accelerate delivery. Yet raw tooling — even impressive AI code generators — is only as powerful as the system that surrounds it. Slow pipelines, flaky tests, unclear ownership, poor onboarding and cognitive overload turn every AI-assisted win into a maintenance headache. The net result: marginal gains in individual tasks but little improvement in organisational throughput.
Friction is the invisible tax that multiplies across teams. It eats developer time, inflates costs and delays revenue-sensitive features. The practical consequence is stark: you can introduce the latest AI assistant, and still see feature cadences stall because the rest of the development system is brittle.
Make the business case: translate hours into pounds and product outcomes
One of the most valuable lessons in the Pragmatic Engineer excerpt is practical guidance on selling DevEx to the C-suite. Executives understand revenue, cost and time-to-market. Translate developer toil into these terms:
- Convert wasted hours into a fully-loaded cost — frame savings as “recaptured productivity dollars” or “free headcount”.
- Quantify hard savings from cloud, vendor consolidation and fewer incidents.
- Show revenue upside: faster feature velocity and higher quality let you capture market opportunities sooner.
Take Etsy as an example. Leadership there committed engineering capacity to DevEx, framed by clear business metrics. That investment turned into better throughput and became part of standard operations rather than a one-off project. That’s the pattern to emulate.
Practical levers to remove developer friction
Reducing friction is not glamorous, but it’s high-leverage. Focus on three categories of work:
- Flow foundations: Reliable CI/CD, fast builds, reproducible environments. These are the rails that let teams iterate.
- Cognitive load: Better documentation, clearer ownership, sensible abstractions. Make decisions visible so engineers can focus on product trade-offs, not infrastructure mysteries.
- Feedback loops: Faster test feedback, meaningful observability and reliable rollback paths. Rapid feedback helps teams learn, not just ship.
Organisations that applied these levers — from large financial firms to consumer platforms — saw dramatic shifts. Capital One, for instance, reduced time-to-market by standardising dev environments and CI/CD practices; the business outcome was faster product launches and measurable market gains.
Measure the right things — especially with AI in the mix
Metrics remain critical, but AI changes what you must capture. Traditional DORA and SPACE metrics still matter — they expose flow and reliability — but AI adds new dimensions:
- Prompting efficiency and validation effort: how many attempts and how much review does an AI suggestion need?
- AI usage patterns: which tasks are delegated and where does AI actually save time?
- Trust calibration: are teams over- or under-trusting AI suggestions?
Second, be pragmatic about causality. Perfect ROI modelling is seductive but often unnecessary. Reasonably credible measurements, tied to a clear narrative, will usually win the day. Instrumentation should be team-level, privacy-aware and designed to answer whether changes produce faster customer value — not just more lines of code.
How to get started this quarter
- Run a short friction audit: measure build times, test flakiness and average wait times for code reviews.
- Pick one high-impact area (CI reliability, onboarding, or test suite speed) and deliver a visible win in 8–12 weeks.
- Frame the win in business terms — reclaimed developer hours, reduced cloud spend, or days shaved off time-to-market.
Teams that follow this pattern can often unlock “free” capacity equivalent to hiring additional engineers — a message that resonates strongly in tight markets.
A final thought
AI is a powerful enabler, but it magnifies both strengths and weaknesses. The smarter route to velocity is not a parade of point tools; it’s a systematic reduction of friction so teams can focus on product outcomes. If you want concrete next steps, start by reading the Pragmatic Engineer excerpt of Frictionless and then translate its metrics guidance into a short, business-oriented pilot. Remove a single, measurable blocker for one team — document the dollars and days saved — and you’ll have the credibility to scale.
Fix the system, not the tool. That is where true velocity — and sustained competitive advantage — lives.
Leave a Reply