NÃO vale a pena ser dev em 2026

Summary

The speaker argues that working specifically as a software developer may no longer be worthwhile in 2026. He defines the role narrowly: taking a predefined ticket, writing code, opening a pull request (PR), and repeating the cycle. In his view, AI is making that work faster and cheaper, weakening the need for teams hired primarily to produce code.

He describes four changes he says he witnessed at a startup:

1. Hiring stopped: The company paused developer recruitment and required existing developers to use paid Claude Code subscriptions.
2. More people began opening PRs: Designers, founders, and managers could make changes directly, reducing developers’ role as intermediaries between business problems and code.
3. Work became an automated loop: An AI agent took tickets from Linear, implemented them, opened PRs, and—after automated checks and review—could merge changes without a person writing or reviewing the code.
4. Technical-only roles were cut: Developers were offered the option to move into areas such as marketing or sales. Those who wanted to remain solely developers were dismissed. The speaker says he chose to leave.

He argues that this pattern could spread beyond startups, though larger companies may adopt it more slowly. He also disputes common reassurances: that AI errors require developers to inspect every line, that companies will always need senior developers because juniors must be trained, and that studying traditional programming fundamentals will remain sufficient.

As an alternative to line-by-line human review, he proposes a system of tests, automated checks, automated review, and monitoring. He says this approach may require a smaller team.

The speaker expects computer-based jobs beyond development—including design, product ownership, and digital advertising—to be affected. He predicts that versatile generalists who can work across product, design, sales, marketing, and technology while using AI may be better positioned.

Advice and resources

• For people considering a developer career: He advises against relying on coding as a long-term job path, especially if writing code is the main attraction.
• For current developers: Learn AI tools and agent-based programming, how to integrate AI into existing systems, and how these systems work. He also recommends building skills in areas such as sales, marketing, and project management.
• Technical knowledge: He recommends developing enough technical intuition to understand databases, front ends, back ends, deployments, and environment variables—not necessarily low-level coding or interview exercises.
• What to deprioritize: He suggests reducing emphasis on detailed code-organization practices such as Clean Code and Domain-Driven Design, arguing that AI agents increasingly write and maintain code.
• Book: Shape Up by Ryan Singer of Basecamp, which he describes as a resource on deciding what to build and estimating how long it will take. He says it is available to read for free on Basecamp’s website.
• Video reference: His video “Stop Checking Code” covers the automated-review idea. This video does not present a step-by-step tutorial.

These are the speaker’s observations and forecasts, rather than a neutral survey of the job market.

Main speaker and source

• Daniel Bergholz — narrator and source of the firsthand startup account, analysis, recommendations, and predictions.