TL;DR
Get the latest gadgets delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
The Pragmatic Engineer’s 2026 industry report says AI coding tools are changing how many software engineers work, with some coordinating five to 10 agent sessions at once. The report also flags concerns about code quality and review practices, while arguing that teams and planning remain important. Its observations come from interviews and conference research, not a representative industry-wide survey.
AI coding tools are changing software work across parts of the tech industry, according to a 2026 report from The Pragmatic Engineer, which draws on interviews, conference observations and access to technology companies. The report says some engineers now coordinate five to 10 coding-agent sessions in parallel, while warning that code quality and review practices have become concerns as AI-generated output grows.
The report’s author says the shift gathered pace after coding models improved near the end of 2025. Many engineers, the report argues, are writing less code by hand and spending more time directing agents, checking results and moving among concurrent tasks. This is presented as an observed trend, not a measured estimate of how all engineers work.
Several practitioners describe parallel workflows. Claude Code creator Boris Cherny said he works across five terminal sessions and runs another five to 10 Claude sessions on the web. Cockroach Labs co-founder Peter Mattis described a typical cognitive limit of about five to 10 concurrent agent sessions, sometimes with subagents. Dima Zaytsev, a software engineer at Linear, said he rotates among five to 10 local worktrees, prompting one agent while checking another’s output.
The report also identifies problems accompanying adoption: assumptions about code output no longer hold, reviews can become performative, and quality and reliability may be falling. Those are the author’s findings and characterizations; the source does not provide a quantified, industry-wide measure of quality decline. It argues that some fundamentals have not changed: teams and planning still matter, even as tools and development practices shift.
AI Changes the Shape of Engineering Work
The shift matters because coding agents can change how engineering time is allocated, not just how code is produced. If developers supervise several agents at once, companies may need to rethink task assignment, review capacity and testing. More generated code does not by itself show that teams are delivering reliable software faster.
The report’s concerns about review and reliability point to a practical challenge: organizations must judge whether their checks can keep pace with the volume and speed of agent-assisted work. The report does not establish that every company is seeing these problems, but it identifies them as issues practitioners should watch as adoption spreads.
For workers, the account suggests that directing tools and evaluating their output are taking a larger role in some engineering jobs. It does not establish how this will affect employment levels, required skills across the sector or long-term productivity. Those outcomes remain open questions rather than settled consequences.
AI coding tools for software engineers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
From Coding Models to Agent Workflows
The author based the report on a keynote at LDX3 in New York, attended by more than 2,000 engineering leaders and senior technical staff, as well as visits to OpenAI and Anthropic and conversations with companies including Ramp and Uber. The author also says they received unpublished data from GitHub, Factory AI and Linear, though the supplied report excerpt does not detail those datasets or their findings.
The report places the current change against earlier shifts such as the internet, smartphones and cloud computing. Martin Fowler, a software engineering veteran, said at The Pragmatic Summit that AI’s impact is larger in magnitude than those earlier changes. That is his assessment, not a quantitative comparison. The report’s central argument is that development practices are changing quickly while established needs for collaboration and planning endure.
“Nothing has hit with the magnitude of AI. This is a whole size difference from anything that we’ve faced before.”
— Martin Fowler, software engineering veteran, speaking at The Pragmatic Summit
As an affiliate, we earn on qualifying purchases.
How Broad Are These Changes?
The report offers examples from engineers and companies but does not establish how common these practices are across the global tech workforce. The excerpt provides no survey sample, adoption rate or comparable productivity data. Its claims about declining quality and theatrical reviews are not accompanied by sector-wide measurements, so the scale and causes of those problems remain unclear.
It is also uncertain whether parallel agent use will become standard, whether it improves delivery outcomes, and how organizations will adapt testing and review. The report forecasts further change, including cloud-based coding agents and new AI infrastructure, but those developments are described as emerging directions rather than guaranteed outcomes.
As an affiliate, we earn on qualifying purchases.
Measuring the Next Phase of AI Coding
The report expects cloud coding agents and supporting infrastructure to develop further, alongside changes in how engineers supervise and assess code. It does not set out a specific timetable or predict which tools or practices will prevail.
The next useful evidence will be clearer data on adoption, software quality, reliability and productivity across different kinds of organizations. Until comparable results are available, the accounts of individual engineers show how work is changing for some practitioners, but not what the shift means for the industry as a whole.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is changing in software engineering in 2026?
The report says some engineers are writing less code by hand and coordinating multiple AI coding agents, then reviewing and testing their output.
How many coding agents do engineers use at once?
Practitioners quoted in the report describe working with about five to 10 concurrent sessions. These are individual accounts, not an industry-wide average.
Does the report prove AI is reducing software quality?
No. The author identifies quality and reliability as concerns, but the supplied material gives no broad quantitative measurement establishing the scale or cause of any decline.
What parts of software development remain important?
The report says teams and planning remain important, even as coding tools and day-to-day workflows change.
Source: rss
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
