AI Infrastructure That Enables Your Squirrels

Most people are still using AI like a chatbot. The better use case is squirrel-enabling infrastructure.

One example: taking raw Zoom transcripts and turning them into searchable local knowledge, structured notes, and task follow-through that lives in your system, not behind a Zoom login. Or Teams. Or Meet. Or whatever.

I wanted something else. I did not want more notes. I needed fewer dropped tasks and less SaaS lock-in.

Like most of you, I rarely get to work on one project straight through the day, concentrating on one successive step at a time. Client meetings, family check-ins, (heck, just choosing to have lunch) has us bouncing between calls, emails, client updates, project milestones, and half-finished threads. Every time we switch contexts, we need a clean way to pick the thread back up.

That has always been the headache for me because I have a self-diagnosed condition that I call ‘ADHSquirrel!‘. So, while I’m pretty good at taking notes (scribbling furiously away to get every acorn of idea for storage). I have not always been great at finding them later.

Harnessing The Squirrels

As work and life got busier, I kept trying to more quickly organize the rote parts of my day to lengthen the creative parts. If I could structure the humdrum, I could harness the combined energy of all the squirrels in my brain toward actual problem-solving.

PARA looked like the right system. I would organize my life into Projects, Areas, Resources, and Archives. I would use Obsidian to keep it all humming along. It is a well-thought-out system and should work swimmingly…in theory.

In practice, the overhead started fighting the work.

• Take the notes.
• File the notes.
• Review the system.
• Update the filing.
• Hunt down the action items.
• Make the donuts.
• Repeat.

“Donuts?” Curse you, squirrels!

Meanwhile, Zoom transcripts molder in nested folders. Tasks get buried. Somebody says, “I’ll send that by Friday,” and two weeks later nobody remembers I was waiting on it. The squirrels get restless. Then they start gnawing at the wiring. Then they blame me for feeding them donuts.

So I built a workflow around Claude Code and Obsidian that fits the way I actually work. I call it CWKBrain.

CWKBrain gets pronounced “QuickBrain.”
CWK is an unconventional abbreviation of C’WicK, if you are curious.
Or ChadWicK, if you like your squirrels fancy. #PinkiesUp

The system takes raw Zoom transcripts, or call notes, and turns them into structured markdown files automatically organized (by PARA principles) in my Obsidian vault with:

• client and project context
• decisions
• action items
• owners
• open questions
• proper filing in the right place

Then a second Claude pass updates my task boards with:

• what is overdue
• what has no due date
• what project has gone stale
• commitments that were spoken but never turned into tasks

That solves my real bottleneck: creating an easily searchable chain of decisions and progress from meeting notes to git commits. Obsidian is great at search, so I can usually find what I need quickly. And if it is not obvious, I can ask Claude Code from a terminal running against the vault.

That is where AI earns its keep. Not by locking me into what Zoom thinks matters from a standalone meeting. Not by giving me a chatbot I did not ask for. By handling the dull, high-volume parsing work and filing it in the place where it can be searched within client and project context.

Now I can keep my attention on decisions, priorities, and follow-through, with the goal of a more focused pack of squirrels.

The system makes sure fewer things fall to floor. The human still decides what matters. The squirrels do squirrel things.

The workflow is based on python scripts and limited ClaudeCode access. It can be cautious when it needs to be: with read-only reports, dry runs, explicit confirmation before anything destructive. That is how automation becomes useful in real life. Not by acting clever. By acting carefully.

This is where AI stops being a toy and starts becoming infrastructure.

I would love to hear how you are using AI in your own automations. And if you or your company are trying to move beyond chatbot use, get in touch.