Gergely Orosz hosted The Pragmatic Summit in San Francisco and attended a 50-person workshop organized by Martin Fowler in Deer Valley, Utah, both focused on the future of software engineering. The resulting article distills six predictions drawn from conversations with industry veterans including Kent Beck, Steve Yegge, Laura Tacho, and Thomas Dohmke. What emerges is not a single thesis but a mosaic of data points, anecdotes, and cautious optimism about how AI is reshaping the profession.
The Data Behind the Hype
The most grounded section of the piece comes from Laura Tacho's keynote, which presented exclusive data from DX showing that 92 percent of developers now use AI coding tools at least once per month. That figure alone marks a staggering adoption curve. But Tacho's data carries a crucial nuance that Orosz rightly foregrounds:
"AI is extremely different in every company because each place has its own problems and its own culture. Organizational performance is multidimensional and these organizations are just going off to different extremes, based on what they were doing before. AI is an accelerator, it's a multiplier, and it is moving organizations in different directions."
The implication is uncomfortable. Healthy organizations get healthier. Dysfunctional ones get worse, faster. Some companies are seeing 50 percent fewer customer-facing incidents while others face twice as many. AI does not fix broken cultures; it amplifies them.
Tacho closed with a comparison to the space race that deserves attention for its sobriety:
"I want to urge you to find that balance between a sense of wonder, aiming for Mars, and a moon colony, but also understanding we need to solve problems here on Earth and we have to live in this reality. So, please stay grounded, stay skeptical, stay human. Most of all, stay pragmatic."
AI-Native Teams and Corporate Friction
Rajeev Rajan, CTO of Atlassian, and Thomas Dohmke, former CEO of GitHub, offered complementary perspectives on what AI-native engineering actually looks like in practice. Rajan described teams at Atlassian where engineers write zero lines of code themselves, orchestrating agents instead, with output sometimes reaching two to five times previous levels.
"Efficiency framing is missing the point, it's more about what you can create now with AI which you could not before."
Dohmke offered a useful reality check. He noted that "AI native" has already become a buzzword drained of specificity:
"There's a lot of BS out there about how all day-to-day tasks are now 'AI native', and using agents for everything. I'm a startup founder: most of the time, I'm still dealing with things like old school HR school systems."
The most revealing exchange came when Rajan admitted to buying a personal laptop because Atlassian's own IT department blocked installation of AI tools on corporate machines. Dohmke pounced on the irony, pointing out that a company that literally sells agile development tools could not move fast enough for its own CTO to experiment with Claude Code.
"To all startup founders: when an investor asks how you're preventing the incumbent from doing the same thing you're doing: just tell them the CTO of Atlassian had to buy a laptop on his own money to start coding!"
It is worth noting, however, that corporate IT restrictions exist for reasons beyond mere bureaucratic inertia. Regulated industries face genuine compliance requirements around data handling and code provenance. The anecdote makes for a crowd-pleasing punchline, but the underlying tension between security constraints and developer velocity is more complex than the laughter suggests.
The Fowler Summit and a Skeptic's Declaration
Nearly 25 years after 17 developers gathered at a Utah ski resort to draft the Agile Manifesto, Fowler convened roughly 50 tech leaders at nearly the same location. The gathering produced no second manifesto, but it did yield a declaration from Beck, Tacho, and Yegge that cuts against the prevailing enthusiasm:
"Organizations are constrained by human and systems-level problems. We remain skeptical of the promise of any technology to improve organizational performance without first addressing human and systems-level constraints. We remain skeptical and we remain human."
That statement amounts to a polite warning. No amount of agentic tooling will rescue an organization that cannot communicate, prioritize, or ship. Orosz reinforces this with his own observation that adoption speed across industries is unprecedented. Companies like John Deere, 3M, and Cisco are all rolling out AI tools, with none apparently willing to wait and see.
Shrinking Teams and the Mid-Level Squeeze
Perhaps the most consequential trend Orosz surfaces is the shrinking of engineering teams. Around 20 engineering leaders at the Fowler summit confirmed the pattern. One head of engineering at a 200-year-old agricultural company framed it starkly:
"We are already seeing the end of two-pizza teams (6-10 people) thanks to AI. Our teams are slowly but surely becoming one-pizza teams (3-4 people) across the business."
Orosz also flags a tension that engineering leaders are discussing privately: mid-career engineers face a quiet crisis. New graduates adopt AI tools instinctively, while senior engineers bring irreplaceable architectural judgment. Engineers in the middle risk being squeezed from both sides.
This is where the article's optimism deserves some gentle pushback. The piece acknowledges shrinking teams as a certainty but treats it mostly as an efficiency gain rather than a labor disruption. When a 200-year-old agriculture company halves its team sizes, those are real jobs disappearing. The article would be stronger if it spent as much time on the human cost of the transition as it does on the productivity gains.
Refactoring and Agile in the Agent Era
The final sections address refactoring and the future of Agile, both topics that gain urgency when most code is agent-generated. Orosz argues that refactoring matters more than ever in an AI-native world because code quality still determines long-term maintainability, regardless of who or what wrote the code.
On Agile, the piece notes an ironic return to Extreme Programming practices, the very methodology that predated the Agile Manifesto by several years. Pair programming, test-driven development, and short feedback loops all map naturally onto human-agent collaboration. Beck, one of the original XP practitioners, was apparently unsurprised.
Bottom Line
Orosz has assembled a valuable snapshot of an industry in rapid transition. The strength of the piece lies in its range of sources and its willingness to let contradictions stand. Tacho's data says AI is an amplifier, not a solution. Dohmke says "AI native" is already overused. Beck, Tacho, and Yegge remind everyone to stay skeptical. And yet the underlying message is clear: the shift is happening faster than any comparable technology adoption in living memory, nobody fully understands where it leads, and the organizations that thrive will be the ones that were healthy before the tools arrived.