Prompt Engineering

If you still think Figma is just a “design tool,” you’re already outdated. Continue reading on Medium »

Here’s what I noticed on LinkedIn the day Figma started enforcing AI credit limits, and what nobody’s saying out loud. Continue reading on Medium »

What I learned from reading all 8 free AI design frameworks, the principles they all share, and a bonus prompt so you can use them easily. Continue reading on Medium »

Khi trí tuệ nhân tạo gỡ bỏ mọi điểm nghẽn trong tư duy và sáng tạo, chúng ta phải nhìn lại từ đầu: "tự mình làm ra" thực sự nghĩa là gì? Hãy thử nghĩ về một tình huống quen thuộc: anh em cần viết một báo cáo, soạn một email phức tạp, hay giải quyết một bài toán kỹ thuật. Trước đây khi gặp một vấn đề kỹ thuật hay dữ liệu dự án, mình thường phải ngồi suy nghĩ, thử sai, sửa đi sửa lại. Bây giờ đôi khi chỉ cần vài dòng prompt cho AI, kết quả có thể xuất hiện trong vài giây. Nhanh hơn, gọn hơn,... Thành tích cá nhân có còn là thành tích khi AI làm phần khó thay bạn?

A step-by-step method to feed AI the right inputs in the right order, without prompt gymnastics. Continue reading on UX Collective »

Last Tuesday, I described a habit tracker app in a text prompt. By Wednesday, it had user authentication, a Supabase database, and a… Continue reading on Medium »

OpenAI acquired Promptfoo, delayed adult mode, retired GPT-5.1, published instruction hierarchy and prompt injection research, Anthropic… Continue reading on Medium »

Agents are not just Smart, but Safe too. Continue reading on Activated Thinker »

Modern work and life generate endless information — tasks, notes, projects, goals, meetings, and ideas. Without a system, everything… Continue reading on Medium »

Continue reading on Medium »

Why some AI-built apps work on the first try Continue reading on Activated Thinker »

If you are building AI applications, particularly autonomous agents that can browse the web or read external documents, you are deeply… Continue reading on Medium »

The seduction of the blank prompt: The cognitive atrophy of instant gratification Continue reading on UX Collective »

The Suite Everyone Should Know Continue reading on AI Advances »

How ChatGPT defends against prompt injection and social engineering by constraining risky actions and protecting sensitive data in agent workflows.

IH-Challenge trains models to prioritize trusted instructions, improving instruction hierarchy, safety steerability, and resistance to prompt injection attacks.