Stop Asking. Start Delegating: How I Actually Use AI On My Site

AI is not a smarter Google

I am convinced most people are using AI in the worst possible way.

They treat it like a slightly magical search bar. Type question. Get answer. Copy. Paste. Forget.

I think that mindset is holding a lot of people back. Developers. Designers. Knowledge workers. Even my baseball kids who ask ChatGPT for homework help.

AI is not a better Q&A machine. It is a delegation machine.

You do not "ask" AI. You give it a job.

This post is me making that shift concrete. I just shipped six AI gallery pages on my site, built entirely around that idea. Not as a gimmick. As infrastructure for how I work, learn, and build.

Why I stopped asking AI questions

The turning point was basically frustration.

My workflow looked like this for months:

• Open ChatGPT
• Ask something like "How do I X in Astro / Svelte / Next"
• Skim the answer
• Try the snippet
• Debug for 30 minutes anyway

The answers were fine. Sometimes even useful. But nothing stuck.

I would ask the same class of questions over and over. Same concepts. Same patterns. Same gotchas. No real accumulation of knowledge. Just one-off transactions.

Then I noticed something: the few times I actually got huge value from AI, I was not asking. I was delegating.

"Rebuild this layout using CSS grid, but keep these class names."

"Refactor this component, keep the same API, and annotate the performance tradeoffs in comments."

"Act like my annoying senior engineer and poke holes in this data model."

That felt different. Less like search. More like a teammate who does legwork while I keep steering.

Delegation > questions

So I made a decision: treat AI like a junior colleague with unlimited patience and questionable taste.

That means:

• I do not ask "How do I do X".
• I say "You are responsible for X. Here is context. Here are constraints. Here is the definition of done."

The shift sounds subtle. It is not.

When you ask a question, the model guesses what you want.

When you delegate a job, you tell it what you want and where it fits inside a bigger system.

I started writing prompts like I write tickets for myself:

• Problem statement
• Existing assets
• Constraints
• Output format
• What I will do next with the result

Suddenly the responses got more predictable. More reusable. Less hallucinated.

And then I hit the second problem.

All those good prompts and workflows were stuck inside random chats. Lost in the scroll.

The missing layer: an AI gallery, not a graveyard of chats

I wanted a place where my best AI workflows could live.

Not as a marketing page. Not as "prompt engineering" porn. As actual, battle tested tools that I come back to.

I tried:

• Notion pages with prompts
• Code snippets in a repo
• Obsidian notes with copy-paste templates

All of that sucked in practice.

When I am in build mode I do not want to context switch into a vault, find the right note, copy, switch window, paste, then adapt the prompt again.

So I did what I usually do: I built something small, opinionated, and slightly selfish.

Six AI gallery pages on my own site.

Each page is a collection of specific delegation flows around one theme. Not generic prompts. Not screenshots. Actual workflows with inputs, constraints, and real outputs from my projects.

Announcing the six AI gallery pages

Here is what I shipped.

These live on my site as proper pages, not hidden tools. They are meant for a broad audience. If you can type, you can use these ideas. If you build things for a living, you will probably adapt them.

1. AI for Web & UI building

This page is basically my front-end assistant in public.

I use AI here as a layout engineer and code janitor:

• "Given this Figma export, generate semantic HTML and a Tailwind scaffold."
• "Refactor this spaghetti CSS into logical layers and BEM-like naming without changing the final rendering."
• "Suggest minimal variants of this component for mobile, but keep the design system tokens intact."

I include real before and afters from my codebase. Not perfect. Not theoretical.

The point is simple. Stop asking "How do I center this" and start delegating "You own the responsive layout, here are constraints, I will review".

2. AI for Content & Writing

I think asking AI "write a blog post about X" is lazy and usually gives trash.

So this page shows how I actually use it to support writing without losing my voice.

• "Take this messy brain dump and map out the structure without rewriting my sentences."
• "Extract all the specific claims from this draft and list which ones need sources."
• "Rewrite only the clunky sentences, keep the informal tone, do not touch any technical terms."

I treat AI as an editor, not a ghostwriter. A fast, opinionless editor that helps me see structure and friction.

The gallery shows side by side drafts and the interventions I kept.

3. AI for Learning & Deep Work

This one is close to my biohacker side.

I use AI as a study partner that never gets bored of my stupid questions but still forces me to think.

• "Act like a coach, not a lecturer. Ask me questions until I can explain React Server Components from first principles."
• "Given this course outline, build a 4-week learning plan that fits into 90 minutes a day and schedules spaced repetition."
• "Generate ten problems that test my understanding of event sourcing, sorted by difficulty, and hide the answers until I ask."

Most people ask AI to explain things. I think that is the wrong direction.

I ask it to interrogate my understanding instead. That is delegation as well. I delegate the role of relentless tutor.

4. AI for Personal Systems & Life Admin

This page is for the boring parts of life that still eat hours.

Things like:

• "Turn this raw weekly log into a short reflection with 3 wins, 3 misses, and 2 experiments for next week."
• "Given my sleep, training, and work logs, suggest schedule tweaks that protect deep work blocks."
• "Summarise this mess of emails and notes into one decision doc with clear options and tradeoffs."

Here I basically treat AI like an operations assistant.

It does not make decisions for me. It compresses reality so I can decide faster.

5. AI for Coaching & Communication

I coach baseball. I also mentor junior devs.

Both groups need clear feedback, simple language, and the right level of challenge.

On this page I show how I use AI to prepare better conversations:

• "Take these scattered notes on a player and turn them into one clear message with one focus area this week."
• "Rewrite this feedback for a 13-year-old who loves hitting but hates fielding. Do not sugarcoat, stay honest."
• "Given this junior's pull request and my inline comments, propose a short Loom script outline that teaches the underlying pattern."

I do not outsource the human part. The empathy and judgment stay mine.

AI helps me say the thing more clearly and consistently.

6. AI for Experiments & Side Projects

The last gallery is basically my playground.

Anything that feels like a small bet goes here:

• "Brainstorm 10 weird ways to visualise this dataset in the browser. Avoid typical chart types."
• "Given this rough idea for a micro SaaS, create three fake landing page variants that test different value props."
• "Help me design an experiment to see if people actually care about feature X. I want to run it this week with no backend changes."

Here the model is more of a chaos generator that I keep on a leash.

I delegate the job of producing options and experiment scaffolds. I keep the job of choosing and executing.

A broad audience, a specific mindset

I built these pages for a broad audience.

You do not need to be a senior engineer. You do not need to understand transformers. You do not need a perfect "stack".

You only need to accept one idea. AI is a delegation tool, not a vending machine for answers.

If you are a developer, that means thinking in systems, not snippets.

If you are a designer, that means delegating the boring layout variants and keeping the taste decisions.

If you are a teacher or coach, that means using AI to structure your thinking, not to replace your expertise.

The gallery pages are opinionated examples of that mindset in action. They are not "best practice". They are my practice.

How I actually built these pages

Because I know some of you care more about the plumbing.

The site runs on a modern static stack with a light API layer. Nothing exotic. I do not think overcomplicating this kind of content is worth it.

The important part is the structure I chose for the galleries:

• Each gallery is a standalone page with a clear theme.
• Every item has: context, the exact prompt or instruction, constraints, sample input, and real output.
• I track which flows I actually reuse with a simple client-side counter. If something never gets reused, it gets cut or reworked.

So the pages evolve with my work. They are not frozen collections.

When I notice I am repeating a manual pattern, I turn it into a delegatable flow and add it to the relevant gallery.

Example:

I kept rewriting similar "refactor this component" prompts in random chats. That friction was my signal. I stopped, abstracted the pattern, gave it a name, wrote it up in the Web & UI gallery, and now I have a reusable workflow.

The process is almost boring:

• Spot repetition.
• Extract the job.
• Document the delegation pattern.
• Publish.
• Iterate when it breaks.

But boring is good. Boring means it is part of my operating system now.

If you take one thing from this

Stop asking AI for fish. Start hiring it as a clumsy intern who never sleeps.

Give it jobs, not trivia questions.

Describe what success looks like. Define constraints. Explain what you will do with the output.

Then build your own small gallery of workflows that actually earn their place in your day.

Mine just happens to be public now: six AI gallery pages that show how I use this stuff across code, content, learning, life, coaching, and experiments.

If that helps you stop treating AI like a search bar and start treating it like a teammate, then the shift is already happening.

The rest is just reps.