Nano Banana is Electric
Discover the implications of Google's Nano Banana image generation models.
Archives
All the articles I've archived.
2025 50
October 50
-
-
When LLMs Meet Robotics, Things Might Get Crazy
Explore how improved large language models could fuse with robotics to transform physical automation and create a new technological era that is fascinating, frightening, and possibly insane.
-
Do LLMs Truly Reason?
Examine the contentious debate around whether LLMs actually reason or just perform sophisticated pattern matching.
-
How Function Calling Works
Function calling lets AI models execute code, query databases, and call APIs. Here's how it actually works and when you need it versus when simple prompting is enough.
-
Most AI Products Are Just API Wrappers. That's Fine.
The AI startup world is full of products that are thin wrappers around OpenAI or Anthropic APIs. That's not a problem. That's how most of the economy works.
-
Rate Limits Are a Feature, Not a Bug. Here's Why.
Rate limits feel like artificial restrictions. They're not. They force you to build better systems with queuing, caching, and backpressure. Here's why unlimited access would actually be worse.
-
When Streaming Actually Matters
Streaming makes AI responses appear faster by showing tokens as they're generated. Here's when it actually improves UX and when it's unnecessary complexity.
-
When NOT to Use AI
Not every problem needs AI. Sometimes a regex is better. Sometimes a database query is better. Here's when to skip the AI and write simple code instead.
-
Probabilistic Versus Logical Reasoning in Generative AI
Compare probabilistic reasoning in LLMs with logical reasoning systems and understand when to use each approach.
-
A History of Semantic Search and LLMs
Trace the evolution of semantic search from 1960s keyword matching to modern LLMs, and understand how these technologies built on each other.
-
Voice Generation is Getting Better and Better
Explore recent advances in AI voice generation and what they mean for content creation, accessibility, and the future of human-AI interaction.
-
Using JSON in Image Prompts
Learn how to use structured JSON prompts for more precise image generation control and consistent results across AI art tools.
-
Are LLMs Bringing Us Closer To a Universal Language That Humans Can't Read?
Explore whether LLM embedding spaces represent a universal language of meaning that transcends human-readable text.
-
LLMs And Language As Numbers
Discover how LLMs transform language into mathematical representations, and why this numeric encoding is key to their understanding and generation capabilities.
-
AI Runs on Rare Earth Metals. China Has Them.
Everyone talks about training models and writing prompts. Nobody talks about the metals that make AI hardware possible. Here's why that matters.
-
America Needs to Win the AI Race
The US and China are racing to lead AI development. This isn't just about tech dominance. It's about what kind of future we get. Here's why it matters.
-
How AI Decides What Products to Recommend (And Why You Should Care)
Ask ChatGPT for tool recommendations and it gives you specific products. How does it decide? And what happens when AI becomes the new search engine for buying decisions?
-
How AI Search Actually Works (And Who's Powering It)
Perplexity and ChatGPT don't search the web themselves. They use APIs. Here's how AI search really works under the hood, and why companies like Noble matter.
-
Let's Speculate About How Lovable.dev Works Under the Hood
Ever wonder how Lovable.dev turns your one-sentence idea into a full app in seconds? Let's speculate about the hidden prompt engineering that makes the magic happen.
-
Are LLMs Just Vector Databases?
Explore the relationship between LLMs and vector databases, and why they're complementary technologies rather than equivalent systems.
-
Prompt Caching Explained (And Why You Should Use It)
You're probably sending the same 50k tokens to the AI every single query. Prompt caching lets you pay once and reuse it. Here's how it works and why nobody uses it.
-
Stop Marketing AI as a Job Killer
Artisan.co's billboards brag about replacing workers. Outset.ai clapped back. Here's why the "AI will take your job" marketing needs to stop.
-
Stop Maxing Out Your Context Window
Everyone brags about 200k token context windows. Nobody talks about how response quality tanks when you fill them. Here's how to use context intelligently.
-
Stop Trying to Make AI Friends Happen
Friend.com raised millions to make AI your best friend. It's not going to work. At least not yet. Here's why.
-
System Prompts vs User Messages (And Why It Matters)
Everyone focuses on what they ask the AI. Nobody focuses on the system prompt that controls how it responds. Here's the message structure you're probably ignoring.
-
Temperature Settings Actually Matter (Here's How)
That temperature slider everyone ignores? It changes everything. Here's what it does, when to use 0.0 vs 0.7 vs 1.0, and why it matters more than your prompt.
-
What Are Workflows? (And Why You Should Care)
Workflows connect your apps so they talk to each other automatically. Here's what they are, when to use them, and which platforms actually work.
-
How is an LLM Different Than Other AI Advancements?
Understand what makes LLMs fundamentally different from previous AI breakthroughs and why they represent a paradigm shift in artificial intelligence.
-
Jailbreaking LLMs
Explore LLM jailbreaking techniques, why they work, and how to build more robust AI systems that resist manipulation.
-
Prompt Injection Explained
Understand prompt injection attacks - how they work, why they're dangerous, and how to protect your AI applications from manipulation.
-
What the Heck is ReAct Prompting?
Understand ReAct prompting - a powerful technique that combines reasoning and acting to help AI solve complex problems through iterative thought-action cycles.
-
Cursor Versus Cline Bot Versus Lovable
Compare three popular AI coding tools - Cursor (IDE), Cline (VS Code extension), and Lovable (web-based builder) - to find which fits your workflow.
-
Why Product-Led Companies Will Dominate in the Age of Claude Code
Explore why product-led companies have a massive advantage in the AI-assisted development era, and how tools like Claude Code amplify their strengths.
-
What is the Model Context Protocol?
Understand the Model Context Protocol (MCP) - Anthropic's open standard for connecting AI models to data sources and tools, enabling contextual AI applications.
-
Codex CLI Versus Claude Code
Compare OpenAI's Codex CLI and Anthropic's Claude Code - two powerful command-line AI assistants with different strengths and use cases.
-
Being Responsible and Using OpenAI's Moderation Endpoint
Learn how to use OpenAI's moderation endpoint to filter harmful content, protect users, and build safer AI applications.
-
Prompting So AI Isn't Your Sycophant
Learn how to prompt AI to provide honest, critical feedback instead of just agreeing with everything you say - get real value, not empty validation.
-
Yes, You Can Vibe Code From Your Phone
Discover how to build and deploy real applications from your phone using Termius, Digital Ocean, and Claude Code - coding truly becomes location-independent.
-
Creating Great Markdown Files For Bulk Claude Code Operations
Learn how to structure markdown files that enable powerful bulk operations with Claude Code, from multi-file edits to complex refactoring tasks.
-
Vibe Coding Versus Vibe Servering
Explore the difference between vibe coding (AI-assisted development) and vibe servering (AI-assisted server management), and why understanding both matters.
-
MetaPrompting is Awesome, You Should Do It
Discover metaprompting - using AI to write better prompts for AI. Learn how this meta-level technique can dramatically improve your prompt engineering workflow.
-
Chain of Thought Prompting 101
Master Chain of Thought prompting - the technique that dramatically improves AI reasoning by making the model show its work step by step.
-
Tokens Explained - The Currency of AI Language Models
Understand tokens - the fundamental units that AI models use to process text, and learn how they impact cost, performance, and prompt design.
-
What is a Hidden Prompt and Why Do Engineers Use Them?
Explore hidden prompts (system prompts) - the invisible instructions that shape AI behavior, and learn why they're essential for building reliable AI applications.
-
Using XML in Prompt Engineering
Discover how XML tags create structure, clarity, and powerful capabilities in AI prompts - from basic organization to advanced multi-step reasoning.
-
Delimiters - What They Are And Why You Should Use Them
Learn how delimiters dramatically improve prompt clarity, prevent prompt injection attacks, and enhance AI response quality.
-
What is RAG? Understanding Retrieval-Augmented Generation
A comprehensive guide to Retrieval-Augmented Generation (RAG) - how it works, why it matters, and how to implement it effectively.
-
What is Few-Shot Versus One-Shot Learning?
Understanding the difference between zero-shot, one-shot, and few-shot learning - essential concepts for effective prompt engineering.
-
What Makes a Great Prompt
A comprehensive guide to crafting effective prompts that consistently produce high-quality AI responses.
-
What is Prompt Engineering?
An introduction to prompt engineering - the emerging discipline of designing effective instructions for AI systems.