AI Tools Articles
Working notes on ai tools, written from real client projects and what I've learned building brands online.
How to Setup a Google Ads MCP Server (and 5 real lessons I learned)
Google Ads is one of the last ad platforms with no managed MCP server. Here is how to build your own locally or on Cloud Run, and the five things that broke on me first.
Why I Run Some AI Locally and Some in the Cloud
Linux is free as long as you don't value your time. The same is true of local AI. Here is the calculation I run before handing anything to a model, and why I keep local models around anyway.
Which AI Model Actually Makes Sense for Running a Business
In my experience, most business owners pay little attention to what model they're running. Instead, they just know "I use ChatGPT" or "I use Claude". But the difference and the nuance between models (even from the same model provider) matters a LOT when you use it to run a business. Here's my two cents.
Why I Build My Own MCP Servers for Google Ads and LinkedIn
An account can burn three days of budget on a broken ad set and nobody finds out until Monday. So I built my own MCP servers for Google Ads, LinkedIn and Sheets. Here is what actually changed.
Meta Muse Review: OpenAI Finally Has a Real B2C Competitor
Meta has spent two years losing the AI race. Muse is the first thing they have shipped that beat every personal assistant I tried to build myself, and it took 90 seconds to set up.
The Best AI Tools for Students: 6 That Are Actually Useful
Most lists of AI tools for students are affiliate spam for products nobody uses twice. These six are the ones I would actually recommend, split by what they are genuinely good at and where they fail.
AI in Education: What Teachers Actually Need to Know
AI in education gets discussed as either a catastrophe or a revolution. The practical reality sits in between, and the teachers handling it best have made three specific decisions.
Can You Build Your Own AI? A Realistic Answer for Businesses
Business owners keep asking whether they can build their own AI. The answer depends entirely on which of four very different things they mean. Here is the honest cost and effort for each.
AI Guardrails: Why the Safest Model Is Often the Least Useful
Heavily restricted AI models frustrate everyone who uses them for real work. But the loosest model is not the right answer for a business either. Here is how I actually think about the tradeoff.
The Real Limits of AI Search Engines in 2026
I argued years ago that AI would make search worse. I was partly right and mostly wrong, and the specific ways I was wrong are the useful part for anyone doing SEO now.
AI Wearables and Customer Data: What Marketers Should Expect
Smart rings and AI wearables collect a category of personal data marketing has never had access to. Most of it will stay out of reach, and the part that does not raises a question worth answering before it arrives.
Opus 5 Is Just Fable 5 but Cheaper, in Both Directions
After a full build session on Opus 5, my verdict is that it's Fable 5 at a fraction of the price. The discount is real, and so is what you give up for it.
The Apple–OpenAI Breakup: What Tech's Messiest Divorce Teaches Marketers
Apple and OpenAI went from the biggest distribution deal in tech to a trade-secrets lawsuit in two years. Here's what the breakup teaches marketers about borrowed distribution.
Sonnet 5 Stress Test: What It's Good At, What It's Not, and How It Compares
A Sonnet 5 stress test across the work I actually do: coding, copywriting, and tool calling, plus cost against Opus 4.8, Sonnet 4.6, GPT-5.5, GLM 5.2, MiniMax M3, and Gemini.
AI Vendor Lock-In Is the Real Risk Right Now: 3 Smart Moves to Make
Anthropic is running Microsoft's old playbook on AI subscriptions, and the bill is coming. Here's what to do before AI vendor lock-in catches everyone flat-footed.
Building Openclaw: One Agent, Many Pipelines, Zero Mondays at a Blank Doc
Why I stopped trying to be consistent on my own and built an agent to run my owned-media operation instead. The boundary between me and Openclaw, and what it actually catches.
The Day My AI's 'Thinking' Ate 97% of the Token Budget
A debugging story about a pipeline that logged success and posted nothing. The culprit: a reasoning model that consumed almost the entire budget on intermediate thinking before the content tokens fired.
GLM 5.2 Review: My Honest Take After Daily Use vs Opus 4.7
A GLM 5.2 review after daily use against Opus 4.7, including migrating Draftly.blog across WordPress, Vite, and Next.js. The closest thing to Claude at any price.
Limit Testing: How One Valorant Habit Taught Me Exactly What AI Is Good and Bad At
Limit testing is a Valorant habit where you push a skill until it breaks. Here's how it taught me piano, writing, and exactly what AI is good and bad at.
Siri AI vs ChatGPT: Why Apple Should Be Scaring OpenAI
Apple's Gemini-powered Siri AI matches paid ChatGPT for most things real people use it for. The Siri AI vs ChatGPT shift will eat the free tier and pressure the $20 paid one too.
Bolt.new Review: 4 Honest Observations from Someone Who Built With It
A bolt.new review from someone who used it to build Draftly.blog. Great for first-time builders, rough for anyone thinking about scale.
Gemini vs ChatGPT: An Honest Review After Daily Use
My honest Gemini vs ChatGPT review after using both every day. Gemini wins on images, memory, and cost. ChatGPT is less lazy on long tasks. Here's who actually wins.
Higgsfield Review: Great for AI Agents, Rough Everywhere Else
An honest Higgsfield review. The MCP makes AI agent image generation easy. The UI is clunky, the checkout is a minefield, and credits vanish fast. Here's the real experience.
Minimax vs GLM: A Retrospective on 3 Months With Chinese AI Models
After 3 months building my openclaw bot, Claude got too expensive. Here's my Minimax vs GLM retrospective: GLM crushes coding, Minimax orchestrates, and how I run them together.
The Best AI Image Generation Workflow I've Found (Gemini + Canva)
My best AI image generation workflow uses Gemini to create the image and Canva Magic Layers to clean it up. Here's the exact process I use to get images that actually look good on a page.
Gemini vs Claude: A Practical Review of Two AI Tools I Pay For
A practical Gemini vs Claude review. Gemini wins images and video. Claude wins vibe coding and 'it just works.' Context is a wash. Here's how I actually split the work.
Marketing That Actually Moves the Needle
Occasional notes on SEO, paid ads, and growth, plus every new post, straight to your inbox. Written for operators, not skimmers.