From Planner Chaos to Focus: What AI Taught Me About Co-Intelligent Work
When the Planner Became a Mountain
Months ago, I started the focus boss agent but never saw it through to publication.
Personal to-do’s started growing beyond what my planner pages can hold. My tasks began spilling into the margins.
I use a physical daily planner to help time block, prioritize, and accomplish daily tasks and wins. I use one planner for both work and personal. This aligns with my minimalist lifestyle.
But the physical planner now resembles an unclimbable mountain each time I stare at it. This isn't working. If anything, it produces worry and anxiety.
Turning to Chat for Help
Yesterday, when I settled in after dinner, I hopped in bed to ideate with ChatGPT (Chat, as my kids call it) on some personal/professional branding ideas. Right before starting my chat, I looked at my planner, as I usually do in the evenings, to plan the next day. Immediately, my planner overwhelmed me. That’s when I decided to switch the focus of that chat session to tackling my ever-growing to-do lists using the help of AI. I asked Chat to play the role of something like a master planner. Role assignment is always my first step when starting a new chat with an AI model. Whenever I start a chat, it is rarely to query the AI like the internet. I use it primarily as a domain expert to help me learn or refine my development in certain areas. By the time I engage in a chat, I have usually gone over the task enough to have a targeted idea of how I want to control it. This instance was no different. After providing the AI with the parameters in which I expected it to operate, I then told it to work step by step with me.
The Step-by-Step Flow
- Step: I would dump all of my personal and work to-do's into the chat.
- Step: It would clean up my dump by correcting any spelling mistakes so that the list was a bit more polished.
- Step: I then asked Chat to rank the lists into two buckets: easy to accomplish, and more time-consuming or, in the case of my personal to-do list, more costly.
I asked it to do this based solely on the task titles. Then, also based on the task titles, I instructed it to draw inferences about the tasks. Next, I told it to use its own judgment as a master planner to apply a numbered ranking of the items it already organized across the four buckets so that I could see the lists in order of easiest to hardest/most expensive.
From Rabbit Holes to Repeatable Systems
After this, I told it to recommend easy apps or web-based tools that I could use as a catch-all for capturing anything that struck me at any time, day or night. It provided a decent list, which I commanded it to refine down to the most privacy-preserving options. We went down a rabbit hole. So I brought us back by choosing Microsoft To Do, already available for work and easy to toggle between accounts to capture anything, regardless of work or personal.
As I go days capturing items, I can export them via a simple copy/paste text list for Chat. I then told Chat that this approach would be my repeatable flow.
Voilà! I had an organized list of tasks I could commit to my planner based on difficulty and costliness. This makes it easier to judge whether I can realistically complete the task, or challenge myself to do it. It helps me prioritize the difficult or costly tasks that are also important.
This approach makes me feel less anxious and reminds me that I can actually handle all of this, which was probably already true. I just needed to visualize it for myself. On a practical level, it helps me keep a cleaner planner and adhere better to my time-blocking approach for getting things done.
Rediscovering Focus Boss
But that wasn’t all for my chat last night. As I started to wrap up, it hit me: focus boss agent. The agent I started a few months after joining Microsoft but never brought to publication. Stuck on some of the configuration details related to what I wanted the agent to do and accomplish, I had paused the project. But now, I realized that using Chat this way for the next 3–4 weeks could let me review our interactions, gather insights on how I used it, and answer clarifying questions.
Once all of this analysis is done, I can leverage this output including a tried and true set of configuration parameters, criteria, instructions, and descriptions that can theoretically power my focus boss agent and turn it into my personal Focus Buddy (I like this name better, but may keep thinking of other options) for work, which was always the ultimate goal.
Like a writing coach, this new buddy would be my Focus Coach, helping me prioritize critical work and easy wins.
What AI Really Taught Me
This whole scenario showed me something important: working co-intelligently with AI isn’t about letting the model run wild, it’s about guiding it with intention. Along the way, I realized a few lessons worth sharing:
Dumping tasks into Chat externalized chaos and made it manageable.
Role assignment and step-by-step instructions turned AI into a true co-pilot, not just a search engine.
Clarity and structure reduced anxiety and made space for action.
Iteration created a repeatable flow I could eventually transform into an agent.
Being an Agent Boss is about control and intention — directing the model to serve your goals.
The real crown jewels here weren’t the prompt and context engineering, although that helped immensely, but the way I connect ideas to tools to use tech for meaningful productivity. AI didn’t replace my planning, it amplified it.
This is what being an Agent Boss looks like in practice, what interacting with, using, and controlling these models for good could look like.