As AI handles coding, traditional tech roles will merge. At Anthropic, PMs, designers, and engineers all code. The future is a generalist "Builder" who can handle multiple disciplines, making role specialization obsolete.
Top engineers are no longer just coding specialists. They are hybrids who cross disciplines—combining product sense, infrastructure knowledge, design skills, and user empathy. AI handles the specialized coding, elevating the value of broad, system-level thinking.
The current AI shift mirrors the invention of the printing press. Just as the press made reading/writing accessible beyond a small scribe class, AI is making software creation accessible to everyone, potentially unlocking a new "Renaissance" of innovation.
The core task of writing code is no longer a significant challenge for AI. The focus is shifting to adjacent tasks and higher-level problem-solving, as demonstrated by Boris Cherny, who hasn't manually written code since November 2024.
Anthropic intentionally under-resources some projects, assigning just one engineer. This pressure forces them to automate their work using Claude (“clodify”) to ship quickly, revealing a counter-intuitive management tactic for accelerating AI adoption and efficiency.
The now-massive Claude Code tool was not an instant success. After its public release, it took many months for the broader user base to understand its value and for adoption to accelerate, showing that even revolutionary products can have a slow burn.
To create a breakthrough AI product, design its capabilities around the projected power of models six months out. This means accepting poor initial performance, but ensures you'll be perfectly positioned when more capable models are released.
To foster breakthrough ideas, companies should initially provide engineers with unrestricted access to the most powerful AI models, ignoring costs. Optimization should only happen after an idea proves its value at scale, as early cost-cutting stifles creativity.
Richard Sutton's "Bitter Lesson" suggests general compute always wins. Applied to LLMs, building complex workflows or fine-tuning yields only temporary gains that the next-generation general model will erase. Always bet on the more general model.
AI is evolving from a coding tool to a proactive product contributor. Claude analyzes user feedback, bug reports, and telemetry to autonomously suggest bug fixes and new features, acting more like a product-aware coworker than a simple code generator.
A new product development principle for AI is to observe the model's "latent demand"—what it attempts to do on its own. Instead of just reacting to user hacks, Anthropic builds tools to facilitate the model's innate tendencies, inverting the traditional user-centric approach.
Anthropic's safety model has three layers: internal alignment, lab evaluations, and real-world observation. Releasing products like Co-work as “research previews” is a deliberate strategy to study agent behavior in unpredictable environments, a crucial step lab settings cannot replicate.