Fun story. Yesterday I had lunch, sat down at my computer with a cup of tea and a cat, and started building. Two hours later, I had moved three separate projects meaningfully forward, including one that had been on our family to-do list for a decade. Then the system ran out of tokens at my user level. So I had to take a break while the tokens reset.
That pause gave me a moment to reflect on what had just happened, and what it actually looks like to work alongside AI as a building partner.
The juggling metaphor
Here is what I have learned about working with an AI coding assistant: three concurrent tasks is my current sweet spot. You give it an instruction, it goes to work, and while it is building, you shift your attention to the next thing. Give that one an instruction, shift again. It is like juggling. Three balls in the air is a rhythm. More than three and I start dropping them. Fewer than that and I'm spending time just waiting.
The key is that the AI does the building while you do the thinking. You are not writing code line by line. You are directing, reviewing, adjusting, and moving between workstreams. It is a fundamentally different kind of productivity. It's a much faster realization of ideas.
What I actually built
In those two hours, I was working on three things simultaneously:
- My home dashboard. My personal development workspace had a functional but rough interface: dark background, plain links. I redesigned it into something I actually enjoy opening. Better organization, clearer navigation, room to grow. It will eventually be the hub for everything: recipe collection, home automation, project tools, all in one place.
- A digital recipe collection. This is the one that had been on our list for a decade. We have handwritten recipe cards from my husband's grandmother, passed down through his mother. Cinnamon buns, borscht, perogies, cabbage rolls. Years of handling the recipe cards every time we make a batch had them worn out and smudged. This was so easy it's kind of funny how long it took to finally finish, except then I have to remember that it wasn't possible to do this even a year ago. What did I actually do? I took photos of those cards, uploaded them, and the AI parsed the handwriting into structured recipes with ingredients, measurements, and step-by-step instructions. The whole thing is organized like a book with chapters, optimized for reading on an iPad propped up on the kitchen counter. It probably took thirty minutes, interspersed with the other projects.
My next step in the learning path will be to turn this into native iOS and Android apps.
- A content management system. I have been thinking about this one for weeks: a tool to help me manage article creation, social media derivatives, and posting schedules for my professional content. The first prototype came together based on a recorded conversation I had that morning about what features were needed. I use Otter.ai for voice to text transcription and copy/paste the transcript into my agent. I've called this project Camsie, and she's going to be my best friend for content building. Camsie not fully polished yet, but it already delivers every core function I need.
Ten year-old to do item, completed in thirty minutes
The recipe book is the one that keeps hitting me. We have talked about digitizing those recipes for years. Every holiday, someone mentions it. Every time we pull out a faded card to make Baba's cinnamon buns, I think, "We really need to get these into a system." And it just never happened, because the effort always seemed disproportionate to the reward.
With AI, the barrier collapsed. Upload a photo of a handwritten card. The system reads the handwriting, structures the data, generates a clean recipe page with a hero image, ingredient list, and numbered instructions. Done. Next card. The collection now has every recipe of importance that our family has today, organized into chapters that can easily grow over time: Baba's Recipes, Breakfast, Desserts, General, Cleaning Supplies, each with its own visual styling.
And if we want to add a new recipe? Take a photo. Upload it. That is the workflow. It does not need to be any fancier than that.
What this means for how we work
I share this to illustrate a shift in how work gets done using new workflows. This is what applied AI actually looks like in practiceI share it because it illustrates something important about where AI tools are right now: they are practical building partners for real projects that real people care about.
Take a look at it:
Baba's recipes are finally preserved! A professional quality tool built for exactly what I need it to do. A home dashboard that will become the digital front door for everything our household runs on.
The skill is not coding. I am not writing code. The skill is knowing what you want, being able to describe it clearly, and being able to evaluate what the AI gives you back. That is the same skill set I coach teams on every day, just applied to building rather than writing or analyzing.
The stopping point is the interesting part
When the tokens ran out and I had to pause, my first instinct was to buy more and keep going. But the pause was actually useful. I had thrown the football far down the field on all three projects. One of them is fully done, but all of them are usable today. The recipe book is complete and live. The content system needs some polish. The dashboard will keep growing as we add more integrations.
That is the real rhythm of working with AI: intense bursts of building, followed by stepping back to assess. Not everything needs to be finished in one sitting. The point is that the distance between idea and working prototype has collapsed to almost nothing.
If you have a project that has been sitting on your to-do list because the effort seemed too large, a digital archive, an internal tool, a workflow you have been doing manually, it might be closer to done than you think. The barriers that kept it there may not exist anymore.