Nolan Lawson admits a complete reversal on LLMs, now using Claude Code for 90% of his coding. He traces this shift from skepticism about hype to recognizing a practical threshold where AI tools became indispensable for busywork, bug detection, and even security audits.

The post critiques the polarized debate, noting how early adopters included crypto speculators and management seeking to replace developers. Lawson argues the core breakthrough has already happened; future gains will come from better tooling and multi-agent systems, not necessarily smarter models, making the technology viable today.

His advice is to experiment and remain curious, as the field is evolving rapidly. He urges developers to avoid tribalism, acknowledging that the future of software work is uncertain but already being reshaped by these tools, making empathy and open-mindedness essential.