TL;DR
Get wellness gear 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
A report based on a keynote to more than 2,000 engineering leaders says AI coding tools are changing how many software engineers work, including wider use of multiple agents in parallel. The account describes faster development workflows alongside concerns about code quality and reviews, but does not provide industry-wide data establishing how common these shifts are.
A 2026 report from The Pragmatic Engineer describes AI coding tools reshaping software development, with some engineers using several agents at once rather than writing code line by line. The account, based on a keynote to more than 2,000 engineering leaders at the LDX3 conference in New York, also flags concerns about code quality and reviews as companies adapt to faster AI-assisted work.
The report’s author says the pace and scale of AI’s effect on technology work exceed those of earlier shifts he has experienced, including the spread of the internet, smartphones and cloud computing. Martin Fowler, an industry veteran quoted in the report, likewise characterized AI as a change of a different magnitude. These are assessments of the shift, not measurements of its effect across the industry.
A central example is the move toward parallel agent sessions. Boris Cherny, identified in the report as the creator of Claude Code, described using several terminal sessions and running five to 10 Claude agents on the web alongside local sessions. Software engineer Dima Zaytsev, now at Linear, said he works across several local worktrees, prompting one agent while checking or reviewing another’s output. These accounts show how some developers are organizing their work; they do not establish a typical industry-wide practice.
The report also lists potential downsides: assumptions about code output have changed, reviews can become performative, and quality and reliability may be falling. It argues that some fundamentals remain, including the need for teams and planning. The source does not quantify these concerns or give a broad comparison of software quality before and after AI coding tools became more widely used.
How AI Changes Engineering Work
AI coding tools could change not only how quickly code is produced but also what engineers spend their time doing: directing agents, checking results and coordinating parallel tasks. If those practices spread, companies may need to adjust development workflows, review expectations and the skills they value in engineering teams.
The tension is that faster output does not automatically mean dependable software. The report’s concerns about quality, reliability and code review point to a practical challenge for employers: they need ways to judge AI-generated work, even as familiar assumptions about who wrote the code and how it was produced change. The source describes emerging concerns, not a measured industry-wide decline.
As an affiliate, we earn on qualifying purchases.
From Coding Tools to Agent Workflows
The report draws on the author’s visit to AI labs including OpenAI and Anthropic, conversations with startups and technology companies, and a keynote at LDX3. It says the author also received unpublished data from GitHub, Factory AI and Linear, but the supplied material does not include the data or explain its methods. The conference audience included engineering leaders, CTOs and senior technical staff.
The account places the shift in coding workflows alongside a broader acceleration in AI capabilities that it says became more pronounced after models improved at coding near the end of 2025. Earlier technology changes also altered software work, the report notes, but it presents the current period as unusually fast-moving. That comparison reflects the author’s view and Fowler’s quoted assessment, rather than a formal study comparing past and present disruptions.
“Nothing has hit with the magnitude of AI. This is a whole size difference from anything that we’ve faced before.”
— Martin Fowler, speaking at The Pragmatic Summit
software development workflow tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
How Widespread Are These Practices?
The report offers examples from individual engineers and a conference presentation, but the supplied material does not establish how many developers now rely on multiple agents or how representative those examples are. It also does not include the unpublished company data mentioned by the author, or provide methods and results that would support industry-wide conclusions.
The extent of any change in software quality, reliability or review effectiveness remains unclear. The report identifies these as concerns but does not give figures, define how they are measured or attribute a broad decline to AI tools. It is also uncertain how quickly the practices described at AI labs and among highly productive developers will spread to other teams.
As an affiliate, we earn on qualifying purchases.
Evidence Needed as Workflows Spread
The report anticipates wider use of cloud-based coding agents and supporting infrastructure, as well as changes to engineering practices. Those are expectations, not confirmed outcomes. The next useful indicators will be published data on how teams use agents, how they review generated code, and whether measures of defects, reliability and delivery time change alongside adoption.
For now, readers should treat the examples as a snapshot of early working practices, not a settled picture of the entire technology industry. The source points to a full 29-minute conference talk and says paid subscribers can access its slides, but it does not set out a date for further industry-wide results.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is the main change described in the report?
The report says some engineers are using AI coding agents to handle tasks in parallel, shifting part of their work from writing code directly to directing and reviewing agent output.
Does the report show that most engineers use several AI agents?
No. It gives examples from particular engineers, but the supplied material does not provide representative survey data showing how common multi-agent workflows are.
Is software quality falling because of AI?
The report raises concerns about quality and reliability, but it provides no industry-wide measurements establishing a decline or proving that AI caused one.
Why might the shift matter to technology companies?
Agent-assisted work may change how teams assign tasks, conduct code reviews and evaluate engineering work. Companies still need reliable ways to check software produced with AI tools.
Source: rss
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
