TL;DR
Get monitors, keyboards and dev 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 2026 engineering-leadership keynote says AI coding tools are rapidly changing software development, with many engineers delegating work to several agents at once. The report also flags weaker code quality and more performative code reviews, while noting that teams and planning remain important. The scale of adoption and its longer-term effects are still developing.
A report from The Pragmatic Engineer, drawing on a 2026 engineering-leadership keynote, describes AI coding agents as rapidly changing how software teams work, with some engineers managing five to 10 agent sessions in parallel. The account also warns that code quality and reliability are under pressure, making the shift consequential for companies adopting AI tools as well as the developers who must check their output.
The report’s author says the keynote at the LDX3 engineering leadership conference in New York was attended by more than 2,000 engineering leaders and senior technical staff. The analysis combines observations from visits to AI labs, conversations with startups and technology companies, and unpublished data from GitHub, Factory AI and Linear. The source does not publish the underlying datasets or quantify how widespread each practice is.
One trend described is a move away from developers writing every line by hand. Instead, some engineers give tasks to multiple agents and switch between sessions while the tools work. Claude Code creator Boris Cherny said he uses five terminal tabs, each with a repository checkout, and runs additional Claude sessions on the web. Linear software engineer Dima Zaytsev described rotating among five to 10 local worktrees and asking agents to handle separate tasks.
The report identifies several problems alongside the productivity changes: assumptions about code output no longer hold, code reviews can become “theatrical,” and quality and reliability are down. These are the author’s observations, not presented as results from a published industry-wide measurement. The report also says some fundamentals have not changed: teams and planning still matter, and it does not claim that non-engineers have broadly begun shipping production code.
AI Agents Put Review Under Pressure
Delegating implementation to several agents can change the engineer’s work from writing code line by line to assigning tasks, checking results and coordinating parallel activity. That may affect how software teams plan work, measure productivity and divide responsibility for defects. The report’s warning about review practices matters because generated code still has to be evaluated before it is relied on.
For organizations, the account points to a tension: faster production of code does not by itself establish that software is more reliable or cheaper to build. The report offers a snapshot rather than a controlled comparison, so it cannot show how much agent use improves output overall. Its practical message is that adoption is moving quickly while established checks on correctness and quality may be struggling to keep pace.
As an affiliate, we earn on qualifying purchases.
A Rapid Shift in Developer Work
The report places the current change alongside earlier shifts that affected software work, including the spread of the internet, mobile computing and cloud services, as well as new languages and frameworks. Its author argues that AI’s present impact is larger and faster-moving than those earlier changes. That is an interpretation of the moment, not a measurable ranking of technology transformations.
Martin Fowler, an industry veteran, made a similar comparison at The Pragmatic Summit, saying AI’s impact was “a whole size difference” from changes the industry had faced before. The report says the current surge became more pronounced after coding models improved toward the end of 2025. It also describes practices at AI labs and highly productive developers as possible early indicators, while not establishing that every company or engineer works this way.
“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
As an affiliate, we earn on qualifying purchases.
How Broad Are These Practices?
The report does not provide enough published data to establish what share of developers have stopped writing code by hand, how common five-to-10-agent workflows are, or whether those patterns extend beyond early adopters and highly productive engineers. Its references to unpublished data do not include figures, methods or comparison baselines in the supplied material.
It is also unclear how much the reported decline in quality and reliability is attributable to AI-generated code, how those outcomes were assessed, and whether the problems are temporary or persistent. The report describes trends and concerns, but does not provide a representative survey or controlled evidence that resolves those questions.
As an affiliate, we earn on qualifying purchases.
The Next Tests for AI Workflows
The report expects cloud-based coding agents and the infrastructure that supports them to develop further, alongside changes in how engineers supervise and verify software work. Those are forecasts from the report, not confirmed outcomes. The next evidence to watch will be whether companies publish clearer measures of agent use, development speed, defect rates and maintenance costs.
For now, the source’s account suggests that engineering teams are adapting their workflows faster than reliable industry-wide conclusions can be drawn. Further reporting, disclosed datasets and results from real production environments will help show whether parallel agents improve software delivery without weakening review and reliability.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is the main development described in the report?
The report says AI coding agents are changing developer workflows, with some engineers managing several agent sessions in parallel. It also raises concerns about code quality, reliability and review practices.
Does the report show that most engineers no longer write code by hand?
No. The author says there are signs that many engineers have reduced hand-written coding, but the supplied material gives no representative statistic establishing how common that is.
Are developers really using five to 10 agents at once?
Some named engineers describe workflows involving five to 10 sessions or worktrees. These are individual examples, not proof that the practice is typical across the industry.
What concerns does the report identify?
It points to pressure on code quality and reliability, and says some code reviews have become “theatrical.” The material does not quantify these problems or establish their causes.
What remains unknown about AI’s effect on software development?
The report does not establish the industry-wide adoption rate or whether AI agents improve productivity and software quality overall. More transparent measurements and production data are needed to assess those effects.
Source: rss
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
