AI, Ashby Engineering, and the future
10 hours ago
- #AI in Software Engineering
- #Future of Work
- #Engineering Culture
- Over half of Ashby's new production code since August 2025 is AI-generated, with no increase in customer issues or regressions in code quality, velocity, or onboarding.
- AI is reducing the cost of code production but emphasizes the increasing importance of human judgment, empathy, and customer understanding for engineers.
- Two ground rules are established: empathy cannot be replaced by AI, and engineers remain responsible for what they ship, requiring increased skepticism and critical thinking.
- Working with LLMs involves two modes: sidekick mode for high-risk tasks with human oversight, and delegate mode for low-risk tasks like prototyping.
- Ashby invests in AI tooling, education, and safety layers (e.g., linters, automated reviews, tests) to ensure high-quality code while reducing human burden.
- Future focus includes optimizing code reviews for high-value aspects like abstraction sensibility and investing in automated verification (e.g., tests, static analysis) as writing becomes cheaper.
- Code quality is more critical than ever, as messy codebases degrade AI-generated output, and customer understanding is now a core engineering skill for effective problem-solving.