You have 1 article left to read this month before you need to register a free LeadDev.com account.
Estimated reading time: 3 minutes
Key takeaways:
The bottleneck has moved upstream. AI makes coding fast – but problem framing, decomposition, and defining the right solution are now the real constraints.
Judgement beats output. As AI generates more code, engineers are valued less for writing it – and more for evaluating, validating, and reasoning about it.
Soft skills are now delivery-critical. Communication, prioritisation, and alignment directly determine whether fast AI-driven teams ship the right thing – or just ship faster mistakes.
AI-driven coding is shifting software engineering priorities, making critical thinking, architecture, and communication more valuable than ever.
The rise of AI usage in software development has reconfigured the skills engineers need to succeed.
As adoption of tools like Codex and Claude soars, AI-generated code is becoming standard practice. Like any new wave in the software industry, this brings with it the need for fresh skills. We just don’t know exactly what those are yet.
According to the State of AI-Driven Software Releases report, only 8 % of respondents see the need for new specialized skills as a major challenge introduced alongside AI coding tools and agents.
Many leaders acknowledge an AI skills gap, yet few organizations report it as a major challenge because there is little consensus on which skills matter most.
AI is pushing the industry to re-examine many of the traditional ways of working. In doing so, it is rediscovering that some fundamentals remain invaluable, Amy Carrillo Cotten, director of customer transformation at Uplevel explained.
“Understanding business needs and user problems? Still golden. Decomposing work into small testable chunks? Still golden. Knowing what should be done vs. what can be done, and knowing when technology is not actually the solution to the problem? Always and forever golden,” she added.
Your inbox, upgraded.
Receive weekly engineering insights to level up your leadership approach.
Coding skill to engineering judgement
“The developer’s role is shifting from writing every line of code to orchestrating systems and guiding multiple AI agents, and that changes which skills matter most,” said Duncan Greenwood, VP of EMEA at GitLab.
Skiller Whale founder Hywel Carver echoed this sentiment, emphasizing that the most important engineering skills lie in the fundamentals, alongside a renewed focus on architecture across teams.
“We see more emphasis placed on architecture skills across the team. More people have to think critically about software structure, as code creation becomes faster,” Carver explained.
The report reflects this shift in priorities. When respondents were asked, “over the next three years, which areas of engineering competence do you think will be most in demand due to increased use of AI?”, critical thinking ranked highest at 18%.
Alongside this, architectural design (15%), domain expertise (12%), systems thinking (11%), and communication skills (11%) were also identified as top-ranked competencies.
“When you can generate code quickly, the bottleneck moves upstream. Can you clearly define the problem? Can you break it into testable units? Can you reason about failure modes? The gap isn’t ‘how do I use Claude’ – it’s problem decomposition and systems thinking,” Jesal Gadhia, co-founder of cora.ai, explained.
More like this
Rise of soft skills
This shift is echoed in broader industry thinking, particularly around the growing importance of judgement, communication, and human-centred capabilities.
“We see more emphasis placed on communication, delivery planning, management, prioritization,” Carver explained.
“Traditional line-by-line implementation is becoming less critical, while architectural thinking, product-minded judgment, and the ability to validate AI outputs across security, infrastructure, and business domains are becoming essential at every level. The valuable human skills, such as creativity, strategic vision, and critical thinking, are being elevated,” Greenwood added.
“The core skills developers already bring to the table will still be vital, including curiosity, critical thinking, and problem solving skills. This ensures AI-generated code solves the right problems, and that teams can diagnose issues or outages using data and telemetry to assist SRE and support colleagues,” said Carly Richmond, developer advocate lead at Elastic.
Time to prepare
If coding with AI becomes the default, the most valuable engineers won’t just be great programmers – they’ll be best at directing, validating, and shaping the outcome. How can developers grow into that role, and how can leaders support them?
“We predict a need for more formal learning – developers will learn less by experience because they don’t write the code themselves. Every developer becomes a reviewer in topics where they will struggle to keep their own expertise high, because they’re no longer practitioners themselves,” Carver said.
“That generally means we need less low-level knowledge – we don’t need developers who can write the code perfectly – but we do need them to read, understand and critically evaluate code. That means more developers asking “is this the right way to do this” and thinking across architectures and codebases than we’ve had previously.”
New York • September 15-16, 2026
Speakers Camille Fournier, Gergely Orosz and Will Larson confirmed 🙌
Greenwood added that leadership plays a crucial role, and while the underlying principles are not new, they take on greater urgency in an AI-driven environment. Leaders must foster psychological safety by creating cultures of trust, conduct open retrospectives, and assert accountability without blame, enabling teams to learn quickly enough to keep up with rapid technological change.
