Blog
What I learn from testing AI tools and building with them.
GPT-6, Claude 5.5, Gemini 4: The October 2026 Model Rush, Explained
Three labs shipped new model families in about a month. What each one costs, what free users get, and where to start.
Open Models in October 2026: DeepSeek, GLM, Kimi, and Mistral Close the Gap
The biggest open models will not fit on your laptop, but their APIs are cheap. What shipped this month and how to pick one.
Your AI Coding Agent Needs a Leash: 3 Warnings From October 2026
A model that tried to send data out, a setup script that switched off the sandbox, and attackers running agents. What to lock down.
Apps Are Opening Their Doors to AI Agents: Apple, Walmart, and GitHub in October 2026
Agents used to click through websites built for people. Now big apps give them official connections, with permissions and spending limits.
Best Free AI Tools, LLMs, and Agent Platforms in 2026
18 free AI tools, LLMs, and agent platforms worth testing, with model names updated in October 2026.
10 Unconventional Simple Tool Ideas That Solve Hidden Problems
10 small tool ideas for problems most people never notice, like losing 3 hours to YouTube without feeling it.
12 AI Prompting Tips for Better ChatGPT and Claude Results
12 prompting tips with example prompts you can copy, from building context to getting the AI to rate its own confidence.
Open-Weight vs Closed AI Models: Which Should You Actually Use?
How I choose between open-weight and closed models: cost, privacy, control, and what I actually ship.
I Tested 7 AI Coding Assistants So You Don't Have To
I gave seven coding assistants the same ugly bug. Three found it fast, two made it worse. Here is the scoreboard.
How to Run an LLM on a Normal Laptop (No GPU, No PhD)
You do not need a $3,000 graphics card. You need the right model, one install, and about ten minutes.
Claude vs ChatGPT vs Gemini: 6 Months of Daily Use, Real Differences
I used all three every day for six months. Here is which one I open first for writing, coding, and live facts.
AI Agents Are Quietly Breaking the Old 'App' Model
A chatbot answers. An agent does the steps. What agents are, where they already help, and what to check first.
The Free AI Stack: Everything I Use Daily That Costs $0
My monthly AI bill is close to zero. The free tools I use every day, and where the free tier stops being enough.
Prompt Engineering Is Dead. Here's What Replaced It
Clever wording matters less now. Context, checking the output, and picking the right model matter more.
Building in Public from Kathmandu: One Year of Shipping AI Projects
One year of shipping AI projects from Kathmandu with no team, no funding, and one laptop.
AI Myths I Tested: 5 Claims, and 4 Were Flat-Out False
I tested five popular AI claims myself. Four of them failed.