Writing
Featured on Medium with +14.7K followers and 5 Million impressions across internet. I write about AI, prompt engineering, and the future of software development.
Stanford Just Killed Prompt Engineering With 8 Words (And I Can't Believe It Worked)
ChatGPT keeps giving you the same boring response? This new technique unlocks 2× more creativity from ANY AI model — no training required…
Read on MediumALL ARTICLES
The Day Google Made Coding Obsolete: How Gemini 3.0 and Antigravity Force Us All to Evolve
Don't just read another "Gemini 3 changes everything" story. See the real numbers, the developer extinction event no one else will admit…
Renewed Claude for Opus 4.6. Cancelled it After GLM-5, Again…
How GLM-5's 77.8% SWE-bench score and $1/M pricing ended my $20/month Claude addiction (with exact prompts)
I Cancelled My ~$200/mo Claude API Subscription, Again.
Why I walked away from Claude's API after the price hikes — and what I switched to instead.
I've Been Recommending DeepSeek & Kimi for Months. Then Anthropic Published This.
A breakdown of the most explosive AI security report of 2026 — and what it honestly means for everyone using Chinese AI tools.
I Spent $200 on Claude Last Month. Then I Found GLM-4.6
How Z.ai's new 355B parameter model delivers enterprise-grade coding at 1/7th the cost — and why embedded engineers like me are switching
The Car Wash Question That Breaks Every AI, And the 2-Word Fix Nobody Talks About
A simple riddle exposes the reasoning gap in every major LLM — and one overlooked technique closes it.
I Read OpenAI's GPT‑5.2 Prompting Guide So You Don't Have To
(And I turned it into copy/paste templates you can use today.)
I Just Cancelled My ElevenLabs Subscription (And You Probably Should Too)
The voice AI landscape shifted overnight — here's why ElevenLabs is no longer worth $22/month.
You're Using Gemini 3 Like It's GPT-4. That's Why It's Failing.
Gemini 3 isn't another chatbot upgrade. It requires a completely different prompting approach — here's what actually works.
Stop Saying “It's Just Next Token Prediction” (You Sound Like a 2023 Tutorial)
The reductive dismissal of LLMs as 'just predicting the next token' misses everything that makes modern AI powerful.