Rust Developer States Fear of Working Without Artificial Intelligence Agents
October 28, 2026 · 3 min read
Also published in Deutsch, Português (Brasil), Español
A software engineer specializing in the Rust programming language has publicly declared that returning to a workflow without artificial intelligence assistance is now psychologically difficult. The developer, writing under the first name Nikita, described a personal evolution through stages of denial, anger, bargaining, depression, and finally acceptance regarding the integration of automated coding tools into daily professional life. The central assertion is that while initial fears concerned being replaced by machines or wasting time verifying their output, the current reality involves an inability to function efficiently without them. This shift represents a fundamental change in how code is authored, tested, and maintained within the industry.
The author noted that several recent projects would never have been initiated without the support of intelligent agents. Specifically, a software library known as a crate was successfully created only because an assistant handled the initial heavy lifting, whereas manual effort alone would have resulted in zero hours of progress and no tangible output. Despite this productivity boost, the need for rigorous human oversight remains. The use of specific data structures, such as a B-tree map chosen for debugging convenience, still requires the developer to explain trade-offs and rationale, tasks that cannot be fully delegated. Experience remains critical for arguing against poor suggestions generated by the software, especially when working in unfamiliar technical domains where the quality of automated advice is uncertain.
Recruitment processes and job descriptions have become the new battleground for defining acceptable tool usage. The writer observed that vacancy postings now fall into three distinct categories regarding technology adoption. The first category consists of organizations claiming to be natively built around artificial intelligence, often listing agent usage as a core requirement while sometimes harboring unrealistic expectations of completing twenty tasks per day. The second category includes companies that clearly define which tools are permitted, what subscriptions the employer pays for, and how these utilities integrate with specific product goals. The third category omits any mention of such tools, leaving candidates to discover during interviews whether external assistants are banned, if only limited internal versions are available, or if a fully integrated environment exists but was simply not advertised alongside other benefits like health insurance.
Ambiguity in corporate policy creates significant friction for prospective employees. A candidate might join a team expecting a seamless interface only to find that connecting the assistant to work services, accessing logs, or retrieving build results requires manual effort. Conversely, some teams have fully configured environments where the agent can access repositories, run checks, and interact with production services, yet this detail is absent from the hiring advertisement. The author emphasized that clarifying who pays for subscriptions, whether the tool can read code repositories, and if clients in outsourcing arrangements consent to such usage are now essential questions to ask before accepting a position. Without these clarifications, a developer risks moving to a new role only to revert to manual processes for a significant portion of their workload.
The reliance on these systems is reshaping fundamental knowledge within the profession. The author recounted a joke made to a colleague about not knowing what a version control client is, highlighting a serious trend where a new generation of programmers may never learn to use traditional command-line interfaces for committing code or viewing changes. Instead, an autonomous agent will locate necessary files and propose actions directly. Through protocols like the Model Context Protocol, these agents can be granted access to diverse operational tools, allowing a single conversation to inspect logs in Elastic, verify builds in Jenkins, commit changes, and update task statuses. This level of integration creates a dependency so profound that working without it feels like a regression, making the availability and configuration of such tools a primary criterion for job selection.