Software Is NOT Dead. It's a Golden Age for Developers

Summary

The presenter argues that software development is not dead. Instead, AI-assisted engineering may be creating new opportunities for developers, DevOps specialists, cloud architects, and startup founders.

Examples behind the argument

• Adobe-style alternatives: The video describes developer Brandon Thomas using Claude Code and Rust to build an open-source project called Craft in roughly 10 days. It reportedly includes seven applications intended to resemble Adobe products, though the subtitles name only Vector Craft, Photo Craft, Film Craft, Live Craft, and Light Craft. The presenter says the result was imperfect—Thomas reportedly estimated its accuracy at around 30–40%—but sees its speed and scope as significant. The video also says HSBC downgraded Adobe from “buy” to “hold,” citing open-source equivalents as one reason.
• Reverse Engineer Anything (REA): The presenter describes REA as a framework that uses AI agents to coordinate existing reverse-engineering tools. The project reportedly gained substantial GitHub attention after new features and integrations were released.

How the technology works

The presenter places these developments in the context of established reverse engineering. Compiled programs turn source code into binaries, and tools such as JAD, strace, ltrace, GDB, IDA, Hopper, and Ghidra have long helped developers inspect or reconstruct how software behaves. The video’s central point is that these tools are not new; what has changed is how an AI agent can plan and run the work.

In the described REA workflow:

1. A user asks how an application or feature works.
2. An AI coding agent plans steps and calls REA and its connected tools, including through MCP integrations.
3. The tools inspect the software and return evidence such as pseudocode, references, confidence levels, and unresolved questions.
4. The agent repeats the process until it has enough information to address the request.

The presenter characterizes this as automating and looping a process that engineers previously carried out manually.

Career and project roadmap

The video offers an opportunity-finding approach rather than a step-by-step coding tutorial:

1. Find a painful, time-consuming task, especially work that is repetitive, slow, or dependent on only a few specialists.
2. Reuse existing tools rather than building everything from scratch.
3. Connect them in an engineering loop around the problem, with an AI agent coordinating the steps.
4. Apply the result within a company, through freelancing, or as a startup idea.

Suggested project areas include an incident-response agent that works with logs, traces, and kubectl; a legacy-system modernization workflow; and a performance-improvement loop that uses profilers and load testers to identify and address issues.

The presenter stresses that this approach is not easy or guaranteed to succeed. A beginner might build a smaller project to demonstrate skills for a job, while an experienced engineer could target broader organizational or commercial problems.

Reviews, guides, and tutorials

• Analysis: The video examines AI-assisted reverse engineering and argues that the main change is the automation and orchestration of existing tools.
• Career and startup guidance: It provides a general roadmap for identifying problems and building tool-based solutions.
• Product review or hands-on tutorial: No detailed product review or build-along tutorial is provided.

Main speaker and sources

• Main speaker: Savinder Puri, presenting the analysis and career guidance.
• Sources discussed: Brandon Thomas and the Craft project; the Reverse Engineer Anything (REA) repository; Claude Code; and reverse-engineering tools including JAD, strace, ltrace, GDB, IDA, Hopper, and Ghidra.