π‘ My Philosophy
- AI tools are the new Stack Overflow β an accelerator for learning and shipping
- They can make developers faster and more productive, but only with deliberate practice
- Like learning a new programming language, thereβs a steep learning curve:
- How to prompt effectively
- When to provide or clear context (e.g., using
/clear to refocus)
- Understanding model limitations and strengths
- Iβm a strong believer and adopter, but also mindful:
- Not βvibe codingβ β everything must be production-grade
- Security-first β careful with data, code sharing, and compliance
β‘ Context
- Driving AI adoption inside engineering at Crystal Intelligence
- Exploring how to leverage AI personally & professionally to enhance productivity, automation, and coding best practices
π Professional Use
- Leading AI adoption program for 40+ engineers:
- Split into 3 phases β pilot, ongoing, steady
- Currently finishing pilot phase: access to GitHub Copilot, Cursor, and Claude Code
- Stimulating engagement and collecting feedback so developers form their own opinions
- Monitoring adoption & productivity with Jellyfish AI adoption features
- Goal: determine the best AI coding tools for Crystal engineers and understand their impact on workflows
π§βπ» Personal Use & Exploration
- Daily user of ChatGPT and Claude for work and personal projects
- Coding journey: started with Windsurf (ex-Codium) β moved to Claude Code
- Heavy use of n8n combined with OpenAI/Claude for automations
- Example: automated billing reports for on-call engineers (incidents β overtime β compensation)
- Researching Lovableβs stack (generated code platform) β experimenting with GPT Engineer (repo)
- Next goal: fine-tuning a model or experimenting with fine-tuned models to understand boundaries & capabilities
π Impact
- Kickstarting a culture of AI-augmented engineering at Crystal Intelligence
- Personally building expertise in AI coding assistants, workflow automation, and model customization