TL;DR
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
A report based on a keynote to engineering leaders argues that AI coding tools are changing how software is produced, with some engineers coordinating several agents instead of writing code by hand. It also identifies concerns about code quality, reliability and the effectiveness of reviews, while saying teams and planning remain important. The account is a snapshot of industry practices, not a comprehensive survey, and does not quantify how widespread these changes are.
A report published by The Pragmatic Engineer says AI coding tools are changing how some software engineers work in 2026, with developers increasingly assigning tasks to multiple agents at once rather than writing every line themselves. The account, based on a keynote to engineering leaders and conversations with practitioners, also flags concerns about code quality, reliability and reviews; it does not provide a representative industry-wide measurement of these trends.
The report’s author says the snapshot draws on a keynote at the LDX3 engineering leadership conference in New York, attended by more than 2,000 engineering leaders and technical staff, as well as visits to AI labs and conversations with startups and technology companies. The author also cites unpublished data from GitHub, Factory AI and Linear, but the provided account does not detail those datasets or their methods.
Several practitioners describe coordinating roughly five to 10 agent sessions in parallel. Claude Code creator Boris Cherny said he uses multiple terminal sessions alongside agents running on Claude Web. Cockroach Labs co-founder Peter Mattis described a similar workload, and Linear software engineer Dima Zaytsev said he rotates among local worktrees while agents work on separate tasks. These are individual accounts, not a measure of typical practice across the industry.
The report groups its observations into changes, problems and continuities. It says coding by hand and the traditional centrality of the IDE are receding for some engineers, while assumptions about generated code have weakened and reviews can become performative. It also identifies declining quality and reliability as concerns. At the same time, its author argues that teams and planning still matter and that AI tools have not made non-engineers broadly responsible for shipping software.
How Agent Work Changes Engineering
If the practices described spread, software teams may spend less time directly producing code and more time setting tasks, checking outputs and coordinating agents. That changes what engineering work looks like, but it does not remove the need for people to understand product requirements, test behavior and take responsibility for releases.
The reported concerns matter because faster code production does not by itself establish that software is dependable. If reviews become a formality or teams cannot keep pace with generated changes, organizations may face more defects or spend additional effort testing and maintaining code. The report identifies these risks but supplies no quantified estimates of their frequency or cost.
Products Worth Considering
Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Black
Apple iPad Air 11-inch (M2), 512GB, Wi-Fi, Space Gray
Apple iPad Air (2025) 13‑inch Wi‑Fi 256 GB – Space Gray (M3 chip)
As an affiliate, we earn on qualifying purchases.
From Coding Models to Agent Workflows
The report places the shift after improvements in coding models near the end of 2025. It compares the pace of change with earlier shifts such as the internet, smartphones and cloud computing, while arguing—through comments from industry veteran Martin Fowler—that AI’s impact feels larger to practitioners. That comparison is an attributed assessment, not a measured ranking of technological changes.
Software development has changed repeatedly through new languages, frameworks and working methods. The report’s distinction is that AI agents can now take on coding tasks directly, allowing one engineer to manage several streams of work. The author expects tools and practices to keep changing, but also says established needs such as team coordination and planning remain.
“Nothing has hit with the magnitude of AI. This is a whole size difference from anything that we’ve faced before.”
— Martin Fowler, software engineer and industry veteran, speaking at The Pragmatic Summit
How Widely These Practices Apply
The report does not establish how many engineers or companies now rely on multiple agents, how often AI-generated code reaches production, or whether the described workflow improves productivity overall. Its examples come from selected practitioners and technology organizations, while the cited unpublished datasets are not presented with methods or results in the supplied account.
The scale of the reported quality and reliability problems is also unclear. The article describes code reviews as becoming “theatrical” and says quality is down, but gives no rates, comparison period or independent audit. It remains unsettled how quickly practices will spread beyond teams that are already heavy users of AI tools.
Signals to Watch in Engineering Teams
The report expects cloud-based coding agents, supporting “harnesses” and new AI infrastructure to develop further. Those are forecasts from the author, not confirmed timelines or announced industry-wide standards. The next meaningful evidence will be more transparent data on how teams use agents and on the effects on code quality, reliability, delivery times and maintenance.
For now, the clearest takeaway is a change in working practice among the engineers quoted: they delegate coding tasks to several agents and switch among outputs. Whether that becomes normal across the industry—and what safeguards teams adopt as it does—remains to be seen.
Key Questions
What is changing in software engineering in 2026?
The report describes some engineers using AI agents to write code and handling several agent sessions at once, rather than coding every task by hand.
Does the report show that most engineers use coding agents?
No. It presents practitioner accounts and a conference-based industry snapshot, but does not provide a representative survey showing how common agent use is.
What risks does the report identify?
It raises concerns about code quality, reliability and reviews. The account does not quantify how often these problems occur or establish their causes.
Do AI coding agents make engineering teams unnecessary?
The report does not make that claim. Its author says teams and planning remain important, even as AI changes how coding tasks are carried out.
What developments does the report expect next?
The author expects cloud coding agents, agent-supporting tools and AI infrastructure to grow. These are forecasts, with no specific rollout dates provided.
Source: rss
