Software Engineers are in two camps when it comes to building new software in 2025.
Why it matters: In one camp, we have engineers using Claude Code with orchestration to automate software engineering, project management, frontend development, and backend coding simultaneously.
- Meanwhile, five languages dominate manual coding.
Between the lines: This corresponds with a study by the World Economic Forum, which found that while AI is expected to displace 85 million jobs in 2025, it will also create 97 million new ones.
- Many of those who work in technology will be replaced by AI, while others will continue to be the "human in the loop."
Zoom in: What we have seen is that Cursor changed VS Code from a human tool to a co-pilot.
- Then Claude Code introduced context to the AI for automating software development.
- Today, agentic workflows can automate almost any task that requires interacting with a computer.
- However, most of these technologies require human approval of the work.
Context: Claude Code can be used within an orchestration framework, such as Tmux Orchestrator.
- This software transforms Claude Code into a self-governing, always-on software team that develops software 24/7, manages check-ins, and autonomously reallocates tasks.
- It can scale infinitely by creating parallel teams within tmux windows—an open-source tool.
What's happening: In this era of automation, several programming languages are quietly dominating key areas, each with its strengths, ecosystems, and niches.
- If you're focusing only on the obvious choices, you may overlook the industry's direction. These languages include:
- Rust: Ideal for building production-grade infrastructure automation tools.
- Go (Golang): Developed by Google, this language simplifies production at a massive scale for infrastructure automation.
- TypeScript: A web-centric language emerging as the de facto standard for connecting services, automating data flows, and creating robust, type-safe webhooks.
- PowerShell: The leading tool for system administrators, automating routine tasks across large arrays of servers.
What's next: Despite automation, humans still need to write code, primarily in Python, which remains the most accessible entry point for many programming tasks.
- The fundamental insight is that the definition of a "power user" has evolved. In 2025, true superpower lies not in mastering a single tool but in having the foresight to choose the right tool for each task.
- This may sometimes require writing a script by hand, while other times it involves delegating tasks to a coding orchestration system.
Go deeper: Want to understand the future of work from someone who has helped shape it?
