Anthropic Engineer Says AI Agents That Operate Computers Could Transform Online Jobs

A senior engineer at Anthropic has warned that a new generation of artificial intelligence systems capable of operating computers like humans could soon redefine the nature of work across internet-based professions. According to the company, the impact may begin reshaping software engineering roles as early as 2026.

The technology represents a shift away from traditional AI tools that merely generate content toward systems that can execute tasks independently across workplace software, potentially transforming how digital work is performed.

A New Era of Agentic AI

The comments came from Boris Cherny, creator of Claude Code at Anthropic, during an interview on Lenny’s Podcast hosted by Lenny Rachitsky. Cherny explained that AI systems capable of taking direct action inside digital environments are improving rapidly. Rather than simply responding to prompts, these systems can perform tasks across workplace tools, complete workflows, and manage operations traditionally handled by humans.

He suggested that this capability could eventually extend to nearly any type of work performed on a computer, while acknowledging that the transition may cause significant disruption for many professionals.

Unlike conventional chatbots that generate text or images, agentic AI functions more like a digital operator. These systems can run terminal commands, edit files, navigate code repositories, analyse data, coordinate multi-step processes, communicate with collaborators, and deploy software within enterprise environments.

Claude Code and Anthropic’s Strategic Direction

Anthropic is widely recognized for its Claude chatbot, but Claude Code represents a broader strategic move toward autonomous AI systems. Built on the company’s Claude model family, the tool is designed to handle development-related tasks such as modifying codebases, analyzing projects, and executing technical workflows.

The company released an updated model suite, Opus 4.6, in early February, further strengthening the system’s capabilities. While Anthropic acknowledges that Claude Code has not yet reached the level of experienced human engineers, Cherny described it as a major step toward making agent-based AI accessible to a wider audience.

He noted that the technology provides many professionals with their first practical experience using autonomous AI systems, offering insight into how such tools may integrate into everyday work processes.

From Passive Assistant to Active Executor

One of the most significant changes introduced by agentic AI is the shift from assistance to execution. Traditional AI systems primarily provide suggestions or generate responses, but agentic AI performs tasks directly.

In software development, this transformation could fundamentally alter the role of engineers. Instead of spending large amounts of time writing and refining code manually, professionals may increasingly focus on designing system architecture, verifying outputs, guiding AI-generated work, and making strategic technical decisions.

In a previous appearance on the Lightcone Podcast, produced by Y Combinator, Cherny suggested that the traditional job title “software engineer” could begin to fade by 2026. Programming itself would remain important, but the nature of the work could shift toward defining objectives and supervising automated systems rather than writing code line by line.

Expanding Influence Across Knowledge Work

The impact of agentic AI is expected to extend beyond engineering into other knowledge-based professions. Product managers could rely on AI agents to analyse user data, identify trends, and generate feature roadmaps. Designers may use autonomous systems to produce prototypes, test user experiences, and iterate on interface designs with minimal manual intervention. Operations teams could deploy AI tools to manage routine workflows, reconcile data, and generate compliance reports.

If AI agents successfully integrate with productivity platforms, analytics tools, and customer service systems, automation could expand across most digitally connected professions, reshaping how organisations structure their work.

Economic and Workforce Implications

The long-term consequences for employment remain uncertain. Cherny acknowledged that society has yet to determine how to manage the broader transition. The ability for individuals to create functional software products through simple natural-language instructions could significantly lower barriers to entry in technology development, enabling more entrepreneurs and small teams to build digital solutions.

At the same time, organisations may reduce staffing needs if AI systems assume a growing share of routine responsibilities. While previous waves of automation displaced certain roles and created new opportunities, agentic AI differs by targeting cognitive and creative tasks typically associated with white-collar employment. This shift could reduce demand for mid-level positions while increasing the need for expertise in AI governance, system oversight, and interdisciplinary coordination.

Regulation, Risk, and Reliability

The rise of autonomous AI systems also introduces regulatory and operational challenges. Organisations deploying these technologies must address concerns related to data privacy, cybersecurity, accountability, and system transparency. Questions surrounding liability remain unresolved, particularly in situations where AI agents execute flawed instructions or introduce security vulnerabilities.

Reliability will play a central role in determining adoption rates. Errors made by AI agents in high-risk environments such as finance, healthcare, or infrastructure could expose organisations to serious consequences. As a result, trust and performance consistency will be as important as technical capability.

Read More: OpenAI Signals Renewed Momentum as ChatGPT Usage Surges Again

Competition in the Autonomous AI Landscape

Anthropic operates within a competitive field of AI developers racing to build increasingly autonomous systems. Ongoing improvements in model performance, declining computing costs, and expanding enterprise adoption suggest that experimentation with agentic AI will accelerate in the near future.

Organisations that adopt these technologies early may benefit from cost reductions and faster product development cycles, potentially driving widespread adoption across industries.

Preparing for an AI-Driven Workplace

Cherny advised professionals to begin experimenting with AI tools and understanding how they function. Familiarity with agentic systems, including their strengths and limitations, may soon become an essential skill across industries.

Rather than eliminating work entirely, agentic AI is more likely to redefine professional roles. The transition may resemble previous technological shifts in which new tools initially enhance productivity before fundamentally transforming how organisations operate and how work itself is defined.