🗞️ AI News of The Week: Introducing ChatGPT Images 2.0
OpenAI has launched ChatGPT Images 2.0, a major upgrade to the image generation experience inside ChatGPT that brings sharper fidelity, better text rendering, and more controllable edits.
Mike’s First Take:
It’s impressive and pretty awesome! It is definitely close or better than Nano Banana which is about as good as it gets right now for Generative AI.
Why this matters:
Higher fidelity by default, meaning less prompt wrestling to get usable visuals
Text in images is finally reliable, so posters, thumbnails, and ads stop looking broken
Better in-place editing lets you tweak one part of an image without regenerating the whole thing
Brings ChatGPT closer to being a one-stop content shop for creators and small businesses
🧠 AI Term of The Week: Tool Calling
Definition:
Tool calling is when an AI model decides on its own to use an external tool, like a search engine, calculator, or API, to complete a task instead of answering from memory alone.
Explain Like I'm 5: It is the difference between a kid guessing what time it is and a kid who walks over and looks at the clock. Tool calling lets the AI walk over and check.
Real examples:
Not tool calling: ChatGPT telling you the weather based on old training data.
Tool calling: ChatGPT checking a live weather API and reporting today's actual forecast
Not tool calling: Claude estimating your calendar from context
Tool calling: Claude opening your Google Calendar, listing today's events, and drafting a reply to the 3pm invite
Why it matters now:
Tool calling is what turns a chatbot into an actual agent that gets work done
It connects AI to your real data and apps, so answers are accurate instead of hallucinated
This is the foundation for every serious AI workflow in 2026, from research assistants to inbox managers
💬 Quote of the Week:
“We have two lives, and the second one begins
when we realize we only have one."
🧰 AI Tool of The Week: Paper (Design)
What is Paper?
Paper connects your teams, agents, code, and data on a single design space built on web standards, so nothing gets lost in translation. Good for landing pages, mockups, UI, etc.
Mike: Paper feels like a cooler and more agentic-friendly alternative to Figma.
What it does:
Lets humans and AI agents work side by side on the same design file
Keeps design handoff tight by eliminating the translation layer between tools
Why it's interesting:
Treats AI agents as first-class collaborators instead of bolt-on features
Web-standards foundation means what you design is much closer to what actually ships
Points at where design tools are heading as agents start doing real production work
🛠 AI Tutorial Of The Week: How To Build an LLM Knowledge Base in Obsidian with Claude Code
Description:
In this tutorial, you will learn how to create your own LLM knowledge base in Obsidian using Claude Code. Inspired by Andrej Karpathy's LLM Wiki concept, this workflow turns Obsidian into a searchable second brain that you can query, expand, and maintain with the help of Claude Code.
🤖 What You'll Learn
Karpathy's LLM Wiki: the concept behind it
Obsidian: set up your vault for an LLM knowledge base
Obsidian Web Clipper: capture articles into your vault
Claude Code: run it inside your Obsidian vault
Knowledge base structure: folders, tags, and links that scale
Querying your notes: ask Claude Code questions across your vault
-----------------------------------------------
Mike Murphy — The AI Handyman 🧰
Helping creators & small businesses turn their content & documents into AI‑powered tools.
📚 Tool Calling
❞ “We have two lives, and the second one begins when we realize we only have one.” – Confucius
🧰 Paper
