It’s a small thing. The episode sits in the queue for three weeks, I know I’m not getting to it, and eventually I swipe it away and pretend that’s fine. It isn’t fine. I subscribed to that show because I wanted what was in it, and deleting it unplayed means I got nothing. So I stopped trying to keep up.

The morning I couldn’t remember the name of an app

A few weeks ago I was driving and listening to an AppStories+ episode where Federico and John went through email clients. They compared a bunch of them. One of them sounded like exactly what I wanted, and I made the mistake every driver makes, which is telling myself I’d remember it later.

I did not remember it later.

So I went to the show page for the episode. It has a list of every app mentioned: Spark, Shortwave, Airmail, SaneBox, MimeStream, Hay. That list told me nothing. Six clients with similar names, no indication of which one was the one that made me perk up in the car, and no way to tell without listening to the episode again.

Then I remembered I had my own note for that episode. I opened it, went to the Products section, and there it was:

Spark — The speakers discuss using Spark for organizing emails and managing folders during specific events like WWDC. Also noted for its CLI capabilities that map the app’s UI hierarchy, offering features beyond a typical Gmail CLI.

That was the one. The CLI part is what caught me in the car. The show page gave me a name. My note gave me the name plus the reason it came up, which is the only part I actually needed.

I didn’t write that note. My Mac did, at nine in the morning, while I was at work.

What actually happens every morning

At 7:00 a.m. a script checks the RSS feeds for the shows I follow and downloads anything new.

At 9:00 a.m. a second script picks up whatever downloaded, transcribes it locally with Whisper, sends the transcript to a language model running on my own machine, and writes a markdown note into my Obsidian vault. Then it deletes the audio file, because I don’t need a 90MB MP3 sitting around once the text exists (as well as I don’t listen to the episode from the computer).

The note has four parts. An overview of the episode in a few sentences. A list of key points. A list of every product, app, book, or tool that the hosts actually discussed, each one linked, each one with a sentence explaining why it came up. I also have the link to the show notes in case the LLM gets something wrong.

Nothing leaves my house. No API key, no subscription, no per-minute transcription bill. If you want the background on getting a model running on your own Mac in the first place, I wrote that up in Running LLMs locally on Apple Silicon. The model runs on hardware I already owned locally. I also do not post or share these anywhere, this is a system built for me and only me, but I’m going to show you what to steal.

Why I trust it

This is the part where I’m supposed to hedge and tell you the summaries are a rough approximation and you should really listen to the episode.

I’m not going to do that, because it isn’t how I use it.

For a show like The Rest Is History, I often have no idea what the topic even is. Tom Holland and Dominic Sandbrook will do six parts on Elizabeth I and the Catholic plots, and I genuinely don’t know going in whether that’s something I want four hours of. The summary tells me what the episode is about in a paragraph. Sometimes that’s enough and I move on. Sometimes it makes me want the whole thing, and I go listen to all four hours.

Either way I got something. Before, an episode I skipped was just gone.

It’s either the summary or nothing. Those are the real options, and once you see it that way the accuracy bar stops being “as good as listening” and starts being “better than deleting it unplayed.”

For the record, the summaries have been right about what was actually said. My complaint list is short and petty. Whisper heard “Claude Code” as “cloud code” in this episode, so the overview claims something launched in February 2025 that didn’t, and the product list picked up the misheard version as if it were a separate app.

I deleted that line. Took about four seconds, because the note is a plain markdown file in a folder I own. That’s the part I’d defend hardest if someone tried to sell me a hosted version of this: when the model gets something wrong, I fix it and it stays fixed.

If you want to build one

The rest of this post is the whole build, and I’ll tell you up front that the architecture is the easy part. It’s three scripts and a scheduler. I could describe it in a paragraph and you’d still lose a Saturday, because none of what breaks is in the design.

It’s in the details. A transcript that repeats one sentence forty times and quietly ruins an hour of processing. A run that transcribes an episode perfectly for 33 minutes and then throws all of it away over a single malformed bracket. A product list where the top three entries are the ads you were trying to skip.

I hit six of those. Every one cost me a morning.

Below is the prompt I’d hand someone to build this from scratch, with all six already written into it as requirements, plus which model you actually need for the summarizing and what to do the first time it breaks anyway. You paste it into a coding agent, answer four questions about your setup, and it writes the pipeline against your folders and your feeds.

At the end of March 2026, I was still selecting a month’s worth of daily journal notes in Apple Notes and moving them into a folder by hand. It was the last remaining chore in a process I had already tried to automate twice.

That little monthly job sent me down the Obsidian rabbit hole. I wanted Claude and Codex to see the same folders I saw and handle the boring file moves with one instruction.

Apple Notes could store my notes. Obsidian gave me plain files an LLM could work on. That was the actual reason I moved.

The monthly job I wanted to stop doing

My daily journal showed me the problem first. In Apple Notes, I started by building the page each morning. Then I made a template and copied it into a new note. Eventually I spent time building and testing a Shortcut to make the daily note for me.

The Shortcut helped, but it did not remove the maintenance job. At the end of every month, I still had to select every daily note and move it into a monthly folder. I would occasionally miss one. Then the journal was not actually where I expected it to be.

In Obsidian, I open the app and it opens today’s note. During my monthly review, I tell Claude or Codex to move the month’s daily notes into the new monthly folder. One sentence replaces the selecting and filing I used to do by hand. I glance at the folder structure afterward and it is where it should be.

That is a small task. It is also the kind of small task that quietly makes a system feel like work. I did not need another clever Shortcut. I wanted the files to be available to an assistant that could handle a bounded, boring job while I decided what mattered.

I wrote about the larger version of this in Agentic Obsidian. This is the less exciting version, and honestly the one I notice more often. A folder gets cleaned up without me selecting notes one by one.

What file access changed

The journal is the clearest example, but it is not the only one. Because the notes are files, I can ask Claude or Codex to group related ideas, sort rough notes into an article draft, or reorganize folders while I sleep. When I come back, the result is sitting in the same folder tree I already use.

I keep the judgment. An LLM can arrange the material I captured and handle the first pass. I still decide what the idea means, whether the article is worth writing and what I am willing to publish under my name.

What matters is that AI can work in the place where the rest of my information already lives. I do not have to copy notes into a chatbot, paste the answer back into another app and then remember where I put it. The source and the result can stay together.

I did not migrate my whole life

I started with three folders: Inbox, Journal and Thoughts. I did not import years of Apple Notes and spend a weekend assigning everything a new home. I set up a daily note and started using it every day.

I think in folders, not tags. I want 00_INBOX at the top and visible areas for Journal, People, Books and Health underneath it. Someone who naturally thinks in tags may build this differently. I want to glance at the sidebar and know whether the filing happened.

My vault has grown, but the logic has not changed. I use numbered sections for the categories I care about. It is still messy on the edges (I haven’t set an icon for my media log yet), and I am fine with that. I can make it prettier when I have time. The important part is getting the data out of my head and into a place where I can find it again.

The numbers are not a productivity method. They keep the sidebar in an order that makes sense to me. They also make the result of an LLM file move easy to check. I do not inspect every note after a monthly cleanup. I glance at the folder tree, see the month where it belongs and move on.

AI access got me interested. Portability made the switch easier to trust. My Apple Notes were tied to an Apple account and awkward for me to export. In Obsidian, I have named folders and plain files. If I change tools later, my archive does not depend on one notes app coming with me.

This fits the rule I used in The Great App Consolidation: What Happens When You Actually Cut Your Stack: each app gets a clear job. Apple Notes had become an easy capture box followed by a manual filing job. Obsidian became the place where my notes lived and where I could ask for help organizing them.

Apple Notes still wins at one thing

Apple Notes still wins the capture moment on an Apple device. I miss that. It gets out of the way fast, even if I still think the yellow is horrible.

Obsidian wins what happens afterward in my setup. Everything lands in Inbox, so I do not have to choose the final folder while the thought is still forming. I sort it later when I have time, or I ask an LLM to do the first pass.

That is the part I would keep even if I used a different notes app. A capture system that demands a folder choice at the moment of an idea is asking too much from someone trying to get out the door, finish a meeting or save a thought before it disappears. Inbox lets me capture first and make the filing decision at a calmer time.

The rule that stopped me from bouncing off Obsidian again

I tried Obsidian at least three times before this stuck. Each time, I made the same mistake: I started installing plugins and building a complicated system before I had used the basic app long enough to know what I needed.

Do not install any plugins on day one. Learn the basic functions. Use a few simple folders. Then limit yourself to one new plugin a week.

Obsidian can become a hobby very quickly. That is useful if the hobby is what you want. It is a bad trade if you only need a place for your notes to land and stay findable.

If I were starting again, my first week would be boring. I would create Inbox, Journal and Thoughts, set up the daily note and stop. I would learn how to create, move, find and link notes before installing anything. When a real problem kept showing up, I would add one plugin that solved it. Then I would wait another week.

My three failed attempts came from making Obsidian complicated before I had a problem worth solving. The version that finally stuck started with one job: open today’s note and give everything else an obvious place to land.

At the end of the month, I open my review, give Claude or Codex one sentence and glance at the folder tree when it finishes. I do not miss selecting every journal note by hand.

I do something a little dumb with my to-do list. If I finish something that was never on it (empty the dishwasher, answer an email, swap the air filter) I’ll write it down after the fact just so I can check it off. The task is already done. I’m not tracking anything. I just want the little click of marking it complete.

I felt slightly ridiculous about this until I read about a guy and his paper clips.

The jar

In 1993, a 23-year-old stockbroker named Trent Dyrsmid started each day at a bank desk in Abbotsford, outside Vancouver, with two jars in front of him. One held 120 paper clips. The other was empty. Every time he finished a sales call he moved one clip to the empty jar, and he kept dialing until the first jar was empty.

Eighteen months later he was bringing in $5 million for the firm. James Clear, who tells this story on his site, says Dyrsmid was making around $75,000 a year by 24. He didn’t use an app or have a streak counter. A dollar of office supplies and a rule he wouldn’t break.

That’s the same click I’m chasing when I log a finished chore just to cross it out. So the concept works, but I want to be careful here, because the internet loves to turn a story like this into “do this one weird trick and your life changes,” and that’s not true. It works for certain things and it’s useless for most.

Why it works when it works

Three reasons, and none of them are complicated.

1. You can see it. The clips pile up on the desk, in real time, whether or not you open your phone. Most habit apps bury your progress behind a lock screen and a tap.

2. Moving the clip is its own small reward. That’s the click I keep coming back to. It’s a tiny hit of “done” that’s separate from whether the call went well or the chore was fun.

3. An empty jar is a finishable goal. “Make more calls” is a wish. “Move all 120 clips” is either done or it isn’t, and you know the second you look.

The part most people get wrong

Here’s the distinction that matters, and it took me a minute to see it clearly. A paper clip is a completed-rep counter. The clip moves because you did something and that’s not the same as measuring something.

I wear a Dexcom because I’m a type 2 diabetic, and my blood sugar sits on my phone all day. It’s useful, but it’s a different tool. The number changes on its own. I didn’t finish a task to make it move. When I first tried to fit that into the paper clip idea as an example for this newsletter, it fell apart. Reading a number is feedback. Moving a clip is a rep. If you blur those two together you’ll try to “paper clip” things that can’t be counted, and you’ll quit when it doesn’t work.

So the test is simple. Can you point at a single, finished action and say “that’s one”? If yes, the jar can help. If you’re measuring a state (mood, weight, energy, how present you were with your kids) the jar has nothing to grab.

Where I actually use it

I’ve never owned the literal jar. That’s the whole magpie move: I stole what works and left the rest. The part worth taking is a counter for finished reps. The part I skipped is a jar of paper clips cluttering my desk. What I have instead is a Todoist project called Recurring, and it does the same job.

Restart the Mac mini once a month, change my razor every two weeks, swap out the air filter. One of them (drain the water heater) only happens once a year, and without the checkbox I would absolutely never remember it exists.Each one is a rep I check off, and the check is the point. None of them are exciting, and that’s exactly why the little completion marker keeps me doing them.

The one that matters most looks a lot like Dyrsmid’s calls. I have a task that repeats every single day: post one Note and reply to a few other writers to slowly grow this newsletter. On its own, any single day of that feels like nothing. Checking it off, day after day, is what keeps me showing up when the results are slow. I want people to learn and grow, by talking with others daily, I’m growing relationships and hopefully helping people out.

This is the same reason I keep my task units tiny, which I wrote about in The 5-Minute Task System for Busy People. A small rep is one you’ll actually finish, and a finished rep is one you get to count. If you want the “do it because it’s easier than avoiding it” version of this, that’s You Don’t Need More Discipline, You Need Less Friction.

The honest caveat

It’s not a catch-all, and I’d be lying if I sold it that way.

It only works for things you can reduce to one clean, countable rep. And there’s a failure mode I’ve walked straight into: the check can become the goal. When I write down a chore I already did just to mark it done, that’s harmless, but push it far enough and you’re optimizing for boxes instead of the actual work. The clip is supposed to follow the rep, not replace it. If you use Todoist, a recurring task is basically a paper clip you never have to buy.

But for the boring, repeatable things you keep dropping? A visible count of finished reps is the cheapest fix I know. Two jars, a fistful of clips, one rule you don’t break. Or a recurring task and the small satisfaction of the check. Just don’t ask it to measure anything.

Disclosure: This article contains affiliate links. If you purchase through them, I may earn a small commission at no extra cost to you.

Six months ago I pulled a 32-billion-parameter AI model onto my Mac Studio and ran it locally for the first time. No API. No subscription. No data leaving my machine (and no little billing meter ticking in the corner). It worked. I now have qwen2.5:32b installed locally. It takes 19GB before runtime overhead, which is not nothing, but it has been worth keeping around.

I’ve been using local models alongside Claude ever since. Honestly, Claude is still better at most tasks. Local models earn their spot because they are free, private and always available for the jobs that do not need the best model.

So why bother?

Because if you have a Mac with Apple Silicon and at least 16GB of unified memory, you can do this today. Here’s what you actually need to know.

Why Apple Silicon changed this

Before Apple Silicon, running a large language model locally required a dedicated NVIDIA GPU, a Windows or Linux machine, and a fair amount of configuration. It was hobbyist territory.

Apple Silicon changed two things. First, unified memory means the CPU and GPU share the same memory pool, which gives large quantized models room to run. You still need headroom for macOS, the model runner, and context/KV cache. Second, Apple Silicon is fast at the matrix math LLMs need.

The result: you can run serious models on a Mac mini. Not the fastest inference in the world, but usable for real work.

What you can run on what hardware

Conservative ranges based on what I’ve tested, plus the current Ollama model sizes. “Usable” means fast enough that you’re not watching a loading bar between sentences.

8GB unified memory (base M1/M2/M3 Mac):

• 1B-4B models are the sane target (Qwen3 4B is listed at 2.5GB)

• 7B-8B models can run, but you need modest context and patience

• Skip 14B+ unless you enjoy proving a point

16GB unified memory:

• 7B-8B models run well (Qwen3 8B is listed at 5.2GB)

• 14B is the upper practical range (Qwen3 14B is listed at 9.3GB)

• Sweet spot for a general-purpose local assistant that does not need heavy coding or long-context work

32GB unified memory:

• 26B-32B models make sense here: Gemma 4 26B is listed at 18GB, Qwen3 30B at 19GB, and Qwen3/Qwen2.5 32B at about 20GB

• Qwen 2.5 32B is my installed baseline right now, not the only current recommendation

• 70B-class models are technically possible, practically frustrating

64GB+ (M2/M3 Max, M2/M3 Ultra):

• 30B/32B models have real breathing room for longer context and normal multitasking

• 70B-class models become plausible if you use quantized builds and accept the slower pace

• This is where local AI starts to feel like a workstation setup instead of a party trick

192GB (Mac Studio M2 Ultra, my setup):

• Most consumer-sized model experiments fit without thinking about RAM first

• I currently have qwen2.5:32b installed locally at 19GB, with room to test larger Qwen3 and Gemma 4 models

• Even huge models are not magic: Qwen3 235B is listed at 142GB before runtime overhead, so storage, context length and heat still matter

The two tools you need

Ollama is the simplest way to get started. It’s a command-line tool (and optional menu bar app) that downloads, manages and serves local models. One command pulls a model and starts a local API server compatible with most AI tools. Free and open source.

ollama pull qwen2.5:32b
ollama run qwen2.5:32b

That’s it. You now have a 32B model running locally.

Open WebUI is a browser-based chat interface that connects to Ollama. If you want something that feels like ChatGPT but runs entirely on your own machine, this is it. Also free and open source.

For running models on Apple Silicon specifically, MLX is Apple’s own machine learning framework and produces faster inference on Apple Silicon than Ollama for some models. The Ternary Bonsai models I covered in a recent post run via MLX.

What local models are actually good for

Local models are not as capable as frontier models. Qwen 2.5 32B is good. Claude Opus is better. The tradeoff you’re making is quality for cost, speed, and privacy.

Where local models earn their place:

Bulk processing is the obvious use. Summarizing 50 podcast transcripts overnight doesn’t require frontier-model quality. It requires something good enough, running for free, without burning through API credits. My podcast pipeline uses Ollama for exactly this. I run transcripts of my favorite podcasts basically for free. While popular podcast apps like Overcast and Pocket Casts now support transcripts, I like having my own that I can ask questions about or point AI models to. For example, I have it check episodes of Mac Power Users for links to products that are mentioned.

Private data is next. Notes you don’t want leaving your machine like old journal entries, personal health data, really anything where you’d rather not send it to a third-party server even with privacy promises. I have Obsidian journal entries and health notes that I do not want to send to Claude or OpenAI. A local model can answer a questions like when I went to the doctor and point me back to the exact file where I wrote about it.

Fast iteration matters when you’re testing prompts, building automation workflows or changing the same thing six times in a row. A local model with no network round trip and no per-call bill is often better than a remote API, even if the quality is lower.

Offline work is boring until it matters. Planes, coffee shops with bad wifi, or just not wanting to depend on an internet connection. While this might not be too relevant for my Mac Studio, if you have a high end MacBook, a portable local AI can be really helpful.

The honest tradeoffs

The quality ceiling is real. Local models top out lower than frontier models. For writing, reasoning and complex instructions, Claude or GPT-4 class models still win. Don’t try to replace your best AI tool with a local one. Use it for the jobs that don’t require the best.

Setup takes a little time. Not complicated, but not instant. Getting Ollama running takes twenty minutes. Getting Open WebUI set up takes another twenty. Getting the right model downloaded for your hardware takes however long your internet connection needs.

And fan noise and heat show up, which is useful if you happen to need a hand warmer. Running a large model at full utilization gets warm and loud. On a Mac mini this is tolerable. On a MacBook Pro it’s noticeable.

Where to start

Local model advice ages fast. Check the latest builds to see what is compatible with your machine.

1. Install Ollama (download the Mac app or the CLI)

2. Run ollama pull qwen3:4b if you have 8GB, ollama pull qwen3:14b if you have 16GB, or ollama pull qwen3:30b if you have 32GB+

3. Run ollama run [model-name] to chat with it in the terminal

4. Install Open WebUI for a browser interface

The model that surprised me most: Qwen 2.5 32B. It handles structured tasks, follows complex instructions, and runs fast enough on my Mac Studio that I forget it’s local. That’s the one I use as my default local model for everything that doesn’t need Claude.

The open-source models I reference in my piece Four Open Models Just Made Frontier AI Free all run locally via Ollama or MLX. That post has the specific commands for each one if you want to run them.

The longer version is more complicated, because Zapier is genuinely better at some things. This is an honest comparison from someone who’s used both in production, not a listicle with made-up scores.

What Each One Is

Zapier is a hosted automation platform. You connect apps, define triggers and actions, and Zapier runs your workflows on their servers. You pay per task (each action that runs counts as one task) on a monthly subscription. There’s a free tier (100 tasks/month, two-step Zaps only), and paid plans start around $30/month (the Starter tier) and scale up quickly if you have volume.

n8n is open-source workflow automation software you can self-host. It has the same basic model of triggers, nodes, connections but runs on infrastructure you control. There’s a cloud-hosted version with pricing similar to Zapier, but most people who choose n8n do it specifically to self-host.

Where Zapier Wins

App coverage. As of this writing, Zapier describes its platform as no-code automation across 9,000+ apps. n8n’s integrations page lists 1,933 integrations, with HTTP and webhook nodes covering most of the rest. If you need to connect two obscure SaaS apps, Zapier probably has a pre-built connector. n8n might require you to build an HTTP request manually.

Setup speed. You can have a working Zapier workflow in fifteen minutes. You can have a working n8n workflow in fifteen minutes too, but that assumes you’ve already set up your n8n instance, which takes longer.

Support and reliability. Zapier is a mature company with SLAs, support tickets, and status pages. If something breaks, there’s someone to call. With self-hosted n8n, you’re the support.

Non-technical users. Zapier’s UI is polished and approachable. n8n is also approachable but more technical-feeling. For automations you’re handing off to someone else, Zapier is usually the better choice.

Where n8n Wins

Cost at any meaningful volume. Zapier’s pricing is built around task tiers. A workflow that processes thousands of items can push you up the pricing ladder fast. n8n’s hosted pricing is based on monthly workflow executions, and the self-hosted community edition has no per-task platform fee. My TANDOM content pipeline processes hundreds of articles per week. On self-hosted n8n, that costs me nothing beyond the Mac mini already running at home.

Code nodes. n8n has a JavaScript code node that lets you write arbitrary code inside a workflow. This is the thing that makes complex transformations possible without building a separate service. I use it constantly with string manipulation, conditional logic, formatting API responses. Zapier has a “Code” step but it’s more limited and locked to Python/JavaScript with restricted libraries.

AI integration depth. n8n has native LangChain-compatible AI nodes, built-in support for dozens of models, and agent workflow patterns. You can build a multi-step AI pipeline (retrieve documents, summarize, classify, write output) all inside one workflow without duct-taping separate tools together. This is where I’ve built my most useful automations, including the assistant system I run my whole vault through.

Privacy and data ownership. Your data doesn’t pass through Zapier’s servers. For workflows that touch personal notes, financial data, or anything sensitive, this matters.

Complexity ceiling. Zapier tops out at moderately complex automations. n8n can handle very complex, branching, looping, multi-agent workflows. If you hit a wall in Zapier, n8n can usually get you further.

What I Actually Use Each For

I don’t use Zapier anymore. I moved everything to n8n and haven’t looked back. But if I were starting fresh with zero self-hosting experience, I’d use Zapier for simple integrations (post to Twitter when I publish a blog, log sales to a spreadsheet) and n8n for anything involving AI, code, or volume.

My n8n setup on the Mac mini runs:

• Podcast feed monitoring and routing

• Periodic reports written to my Obsidian vault

None of that would be cost-effective at Zapier’s pricing. All of it runs reliably and for free.

The Actual Recommendation

Start with Zapier if: you need something working today, you’re connecting common SaaS apps, you’re not technical, or you’re building for someone else who needs to manage it.

Start with n8n if: you’re technical enough to run a server or a Docker container, you’re processing significant volume, you want to build AI workflows, or the recurring cost of Zapier bothers you.

Self-host n8n on a Mac mini if you have one sitting around. That’s what I do. It runs continuously in the background and costs nothing. Setup takes about two hours if you follow the official docs.

The automation category is one of the few places where the open-source, build-it-yourself option is genuinely competitive with the paid one.

If you want a working starting point, I packaged the content pipeline I use for The TANDOM as an n8n template. It includes RSS ingestion, dedupe, an AI editor gate, an AI writer prompt, and a WordPress publishing pattern. It is not a one-click business. It is a working workflow you can import, study, and adapt.

Get the AI Content Pipeline template

This post links to my own product; I earn product revenue from it.

I found the weekly review template while looking for something else.

It was too good for its own good. Six sections. Reflection questions. A this-week / next-week split. A scoring system I probably thought would reveal my “real priorities” if I was honest enough with it.

I built it eighteen months ago. I have used it only four times!

The stupid part is that I still feel it sitting there. Not every second. Not dramatically. More like a tab left open in the back of my head. I feel it on Sunday afternoon when the week starts closing in. I feel it when I could “reset the system” but what I want is a shower, food, and silence. I feel it when I open Obsidian and see the template sitting there like a polite accusation.

Nobody asked me to make that template. Nobody is waiting for it. My life will not collapse if I never fill it out again. And yet the undone thing still has weight.

That is a phantom obligation: the guilt of a promise nobody assigned you, including the version of you currently living with it.

The Contract You Forgot You Signed

You read something useful. You watch somebody explain their weekly planning routine. You have a rare clean hour and think, “This would solve a lot of problems if I did it every week.” So you build the template, create the recurring task, make the folder, and write the rule into your system while you are clear, optimistic, and slightly dishonest about your future energy.

Then real life comes back. The kids need something. Work spills over. You sleep badly. Sunday disappears. The task rolls forward. Nothing explodes, so you do not fix it. You also do not revoke it.

That is the part that creates the weight.

A missed task is not the problem. An unrevoked commitment is.

Systems Turn Ideas Into Moral Debt

A thought in your head can fade. A note in your system looks official. A recurring task looks like a commitment. A template looks like a standard you are failing to meet.

I have done this everywhere: Obsidian templates, review checklists, capture processes, habit logs, task-manager rituals, morning routines, and evening routines. The individual pieces are reasonable. The total system is fiction. Stack enough reasonable commitments on top of a real life and you build a fake life beside it.

That is why I keep coming back to friction. I wrote about this from another angle in You Don’t Need More Discipline, You Need Less Friction. A lot of failure is not a character problem. It is a system fit problem.

Phantom obligation is what happens when the system does not fit, but the promise stays alive.

The Most Dangerous Word Is “Should”

The obvious ones are easy: unused templates, recurring tasks you keep rescheduling, trackers with a perfect first three days and nothing after that.

The harder ones hide behind “should.” I should review my week. I should process these notes. I should keep my task manager cleaner. I should use the system I already built.

Sometimes “should” points at something real. But most of the time, in my systems, it means an old commitment is trying to survive without being questioned.

The better question is, “Am I actually choosing this now?”

If the answer is no, then keeping it half-alive is not discipline. It is clutter.

Do a Release Audit

The fix is not to delete your whole system and start over. That is just another dramatic productivity loop, and I have already ridden that one plenty of times. The fix is to audit the promises.

Open the places where your systems live: your task manager, notes app, templates folder, calendar, habit tracker, and saved routines. Do not look for mess. Look for old contracts.

For each one, ask four questions:

1. Did I choose this, or did I inherit it from a moment of ambition?

2. Have I actually used it in the last month?

3. Does it fit the time and energy I really have?

4. If I saw this for the first time today, would I add it?

That fourth question is brutal. It cuts through nostalgia fast.

If the answer is no, release it clearly. Archive the template. Remove the recurring task. Move the habit tracker out of sight. Rename the folder so it stops pretending to be active.

Do not leave it in the system as a little monument to the person you thought you were going to become.

A Magpie system should stay useful because it keeps stealing what works and dropping what does not. I wrote the bigger version of that principle in Steal What Works, Leave the Rest. The “leave the rest” part matters just as much as the stealing.

Paid member resource

I made a companion Phantom Obligation Release Checklist for paid subscribers. It includes a keep/shrink/archive/drop worksheet, release-note template, cleanup checklist, Claude prompt, and worked example for putting down self-assigned commitments that no longer fit. Use it to decide what to keep, shrink, archive, or drop.

Get the release checklist

Shrink the Promise Before You Keep It

Some obligations should not be dropped. They should be made honest.

My weekly review template was too big. It assumed a Sunday version of me with quiet time, open space, and a noble interest in reflection. That guy appears sometimes. He is not reliable enough to build a system around.

The smaller version is a ten-minute scan:

• What is already fixed on the calendar?

• What has to happen this week?

• What is stuck?

• What can be dropped?

That is less impressive than the template. It also has a chance of happening.

This is the move I trust now: before I recommit to anything, I shrink it until it fits a bad week. If it only works during a clean, quiet, high-energy week, it is not a system that works, it’s just a fantasy with checkboxes.

Put It Down on Purpose

The real relief comes from making the release official. Not dramatic. Not ceremonial. Just explicit.

“I am not doing this weekly review template anymore.”
“This routine was a good idea, but it does not fit my life right now.”

Write that down if you need to. Add a small note to the archived file. Change the task title to include “retired.” Move the template somewhere it cannot keep staring at you.

The point is simple: keep ambition current.

The weight was not assigned to you. You picked it up and forgot to put it back down. You are allowed to look at it now and say, clearly, “No, I am not carrying this anymore.”

Paid member resource

I made a companion checklist for paid subscribers. It includes a keep/shrink/archive/drop worksheet, release-note template, cleanup checklist, Claude prompt, and worked example for putting down self-assigned commitments that no longer fit.

Use it to turn the guilt into a decision.

Get the release checklist.

A checklist for finding and releasing commitments you created for yourself but no longer actually serve. Use it when a template, workflow, habit, recurring task, or project keeps creating background guilt even though nobody else is waiting on it.

A commitment that only exists in your head still needs a decision. If you do not decide, it becom…

Read more

For two years I’ve kept Notion open on my Mac Studio. Not because I love Notion. Because I needed somewhere to put structured data. A table of books with status columns, a simple project tracker with dates, a list of apps I’ve tried with ratings. Obsidian is where I think. Notion is where I organize things that need columns.

That split has always bothered me. Two tools means two mental contexts, two syncing systems, two places something might be. So when Obsidian introduced Bases publicly in version 1.9.10, I paid attention.

What Bases Actually Is

Bases adds database-like views directly inside your Obsidian vault. You can create a table, a card view, or a list view that pulls from your notes using filters and sorts. If you’ve used Notion’s database feature, the concept is familiar. The key difference is that Bases works on your actual markdown files. It’s reading properties from your notes’ YAML frontmatter and surfacing them in a structured view.

In practice this means you can create a file called Books.base, point it at your /Books folder, and instantly get a filterable, sortable table of every book note you’ve created: title, author, status, rating, whatever properties you’ve been adding to your notes.

Reorganize your vault without relying on an external database or a separate app.

Where It Works Well

For personal use cases, Bases is genuinely good. I set up three views in the first week:

A reading tracker. My book notes already had frontmatter properties for author, status, and rating. Bases surfaced them as a clean table in about three minutes. I can filter by status (reading / finished / abandoned) and sort by rating.

A content pipeline. Each article draft in my folder has a status property. Now I have one view that shows everything in draft, in review, or scheduled for the week. No more opening five folders to remember where things stand.

An app log. I keep notes on apps I’ve tried. Bases turned this into a searchable grid with columns for category, verdict, and whether I’m still using it.

For these kinds of personal organization tasks (the ones where you want to see your notes in a grid instead of a list) Bases works cleanly.

Where It Falls Short

This is not Notion. Not yet.

There are no relations between databases. In Notion, you can link a project record to a list of tasks, and the tasks know which project they belong to. In Bases, each view is a flat surface. You can filter and sort, but you can’t create relationships between different sets of notes.

There are no Notion-style relational rollups. Bases has formulas and summaries inside a view, but you can’t create a formula that counts tasks in a linked database or sums values across related records.

There’s no collaboration. If you’re working with other people, Bases doesn’t help as this is just a personal-use feature and the views are still early. The filtering is functional but limited, and there are edge cases where Bases doesn’t pick up frontmatter the way you’d expect.

Whether It Closes the Gap

For me, it closes about 80% of it.

The use cases that kept Notion open in my dock were personal organization tasks: tracking reading, managing content, logging things with status fields. Bases handles all of those. I deleted my Notion workspace last week and I haven’t missed it.

If you need relational databases, collaborative docs, or Notion’s web publishing features, Bases is not a replacement. But if you’re using Notion as a personal organizer while keeping your actual thinking in Obsidian, you might be surprised how much you can move back.

The one-brain-one-tool ideal I’ve been chasing for years is finally possible. That’s worth paying attention to.

If Bases makes you want to move more of your system into Obsidian, The Magpie Vault is the starter structure I use for that: folders, templates, assistant instructions, skills, and memory files for turning a vault into a working AI assistant system.

Disclosure: This article contains affiliate or product links. If you purchase through them, I may earn a small commission or product revenue at no extra cost to you.

For about a year I told myself the same story every morning. I needed to be more disciplined. More consistent. The kind of person who just does the thing. I read the books, I made the promises, and I kept landing in the same place: doing the thing for four days and then quietly not doing it on the fifth.

The story was wrong. The problem was never my discipline. It was that I had designed my days so the right thing was hard to start and the wrong thing was easy.

Here is the reframe that actually changed something: most of what we call a willpower problem is a friction problem wearing a costume. You don’t rise to the level of your motivation. You fall to the level of how easy or hard you made the next step.

Friction is the real variable

Think about the last habit you dropped. You probably didn’t decide to quit. You just hit a small point of resistance over and over until stopping was easier than continuing.

The gym bag was in the other room. The app made you log in again. The document was three folders deep. The healthy food needed prep and the junk was already on the counter. None of those are willpower failures. They are friction, and friction compounds. Every extra step between you and the action is a tiny tax, and you pay it every single time.

The flip side is the part people skip. Friction works in both directions. If you can add resistance to the things you want to do less, the same mechanism that breaks good habits will break bad ones. The phone in another room. The app logged out. The snack in the basement instead of the pantry. You are not fighting yourself. You are editing the environment so the easy path and the right path are the same path.

The three points worth removing this week

You don’t need to redesign your whole life. You need to find the specific spots where you keep stalling and remove the resistance there. Start with three.

The starting step. Most things don’t fail in the middle. They fail at the start. The task that’s too big to begin is the one that sits untouched for a week. So shrink the entry point until it’s almost embarrassing. Not “write the article,” but “open the file and type one sentence.” Not “go for a run,” but “put the shoes on.” (I wrote a whole piece on this idea, The 5-Minute Task System for Busy People, because shrinking the starting step is the single highest-leverage move I’ve found.)

The setup tax. This is the work you have to do before the work. Finding the file, charging the thing, clearing the space, logging back in. Pay this cost once, in advance, so you don’t pay it every time. Lay the clothes out the night before. Pin the document. Leave the guitar on the stand instead of in the case. The goal is that when the moment comes, there is nothing between you and starting.

The decision. Every choice you have to make is a chance to choose wrong, or to choose nothing at all. So remove the choice. Same breakfast every day. Same time for the hard task. A default that runs without asking. I’m forgetful by nature, so I’ve stopped trusting myself to remember or decide in the moment. I build the decision into the system once and let it carry me. (That instinct runs through how I set up everything, including my whole scavenged productivity system.)

Why this beats trying harder

Discipline is a finite resource and a bad plan. It asks you to win the same fight every day, forever, on willpower alone. Some days you’ll have it. Plenty of days you won’t, and those are exactly the days the habit needs to survive.

Friction design is different because you only have to win once. You move the gym bag one time. You set the default one time. You shrink the starting step one time. After that, the easy choice and the right choice point in the same direction, and you stop spending energy on the fight entirely.

I even do this with my hands. I keep my keyboard shortcuts on the left side so my right hand never has to leave the mouse. Tiny thing. But it removes a hundred small moments of friction a day.

This isn’t perfect and it isn’t magic. Some habits really do need grit, and some friction is hard to remove. But before you decide you’re lazy or undisciplined, look at the path you’re actually walking and ask a cheaper question first: what would I have to change so that doing this took one less step?

Most of the time, that’s the whole fix. Not more discipline. Less friction.

Four places and none of them were reliable.

I knew what the problem was. I’d known for months. No single app was broken. I’d just accumulated a stack instead of choosing one. Each app had a logical reason for being there. Together they were a productivity liability.

Two of those apps were paid subscriptions doing overlapping jobs. That’s the part that should sting if you recognize yourself here: I was paying real money every month for the privilege of being confused about where my tasks lived. And the money was the smaller cost. The bigger one was time, two years of tweaking and re-tweaking systems instead of just using one.

The Audit

The consolidation started with an audit I’d been avoiding. I opened every app in my dock and asked three questions:

1. What is this for?

2. Do I check it every day?

3. Does anything else in my stack do the same job?

The answers were uncomfortable and obvious. I had two apps that both captured tasks. I had three apps that all held reference notes of some kind. I had two calendars that were supposed to sync and regularly didn’t.

The rule I set before starting: if two tools overlap in function and I can do the job in one of them without meaningful loss, one of them goes. No exceptions for “but I like the design” or “I’ve used this for three years.”

The Decisions (and the Grief)

TickTick was the hard one. I’d used it for a long time. The habit tracking feature was genuinely good. The calendar integration was smooth. But I was already paying for Todoist, and maintaining two task managers was adding switches without adding value. TickTick went.

I consolidated note types into Obsidian. Craft stayed for documents and writing, but reference notes, daily logs, and thinking work all moved to the vault.

The “stuff to do” note in Craft got deleted. The journal task section got deleted. One task manager, one source of truth for what I’m doing today.

You do feel something when you close an app you’ve opened every day for two years. It passed faster than I expected.

What the After-State Feels Like

What changed is the number of decisions I make before I start working. Before the consolidation, the first ten minutes of my day went to checking four places and reconstructing what I was supposed to do. Now I open one thing and the answer is there.

I also stopped losing things. I was never losing important things before, but there was a low-grade anxiety about which system was current, which one I’d updated last, whether the capture I’d made on my phone had synced to the version on my Mac. That anxiety is gone.

The stack I landed on: Todoist for tasks, Obsidian for notes and thinking, Craft for writing and documents, Google Calendar for time. Four tools, clear lanes, no overlap.

Staying Cut

Here’s the part most “I simplified my stack” posts skip. Cutting the apps was a weekend. Staying cut is the actual work, because in three months something shiny launches, gets a glowing review, and the old itch comes back. I’ve reinstalled TickTick once already since this. Lasted four days.

What keeps me from re-cluttering now is a single rule: a new app doesn’t get added, it has to replace. If I want to try something, it has to take over a job one of my four tools is already doing, and the old one comes off the phone the same day. No “let me run both for a while to compare.” Running both for a while is exactly how I ended up with four task inboxes in the first place.

The shiny new app isn’t the threat. The threat is keeping it and the old one.

The Three Questions for Your Own Audit

If you want to run the same process:

1. List every app you opened in the last week.

2. For each one: what specific job does it do that nothing else in your stack does?

3. For every overlap you find: which one would you choose if you could only keep one?

The answers tell you exactly what to cut. The audit is easy. The hard part is trusting what it tells you and actually closing the app. Let me know which apps you decided to cut!

Disclosure: This article contains affiliate links. If you purchase through them, I may earn a small commission at no extra cost to you.

I decided to find out which.

What Setapp Is

Setapp is a subscription service for Mac apps. You pay a monthly or annual fee and get access to a catalog of Mac and iOS apps. As of June 2026, Setapp lists the standard Mac plan at $15/month and an annual plan of either $19/month or $23/month if you need Setapp on multiple Macs. Annual rates are cheaper and there is an option to add more AI (skip this step as you won’t need the AI apps to start). You don’t own the apps, you’re licensed to use them as long as you’re subscribed. Stop paying, lose access.

The pitch is clear: instead of buying individual apps one at a time, you pay one flat fee and get access to a large catalog. If you use more than two or three apps from the catalog regularly, the math can work in your favor.

Apps I Actually Use from Setapp

This is the honest version and not the catalog browse, but the apps that have genuine daily or weekly presence in my workflow.

CleanMyMac. Mac maintenance, malware scanning, storage cleanup. I use it monthly. This is exactly the kind of utility I would otherwise put off buying until I needed it.

Bartender. Hides and organizes the Mac menu bar. This is the kind of app that, once you use it, you can’t go back.

iStatMenus. System monitoring in the menu bar. CPU, RAM, network, disk usage, temperature. I check this constantly with a quick glance up to the menu bar.

Mosaic. Window management. Clean and simple. Replaced a more complex tiling tool I was fighting with.

Timing. Automatic time tracking. It runs in the background and logs how long I spend in each app and on each project. I use this to review my week and make sure I’m actually spending time on what I think I’m spending time on.

PDF Squeezer. Compresses PDFs. Sounds niche. I use it more than I expected.

Apps I Tried and Abandoned

Ulysses. I used it for writing for a while. It’s a beautiful app but I choose to do my writing in Obsidian as it is much easier to write in markdown and have plugins to enhance the experience.

Archiver. For managing archives of old files. Good app, just not a workflow I maintained.

Various menu bar utilities. There are a lot of these in Setapp and I’ve tried most of them. I keep one or two. The rest are additions I didn’t sustain.

The Honest Math

The apps I use regularly from Setapp, CleanMyMac, Bartender, iStatMenus, Timing, would cost real money to buy or subscribe to separately if I rebuilt the stack outside Setapp. The math shifts as plans and individual app pricing change, so I do not treat Setapp as automatically worth it. I treat it as a bundle that has to earn its place.

The real question fro me is whether the optionality is worth paying for. Setapp’s value isn’t just the apps you use, it’s the ability to try apps without committing to a purchase. I’ve discovered things I now rely on by downloading them with no risk. That has real value, even if it’s hard to quantify or justify to myself.

Who Should Subscribe

Setapp makes sense if you’re a Mac power user who regularly evaluates new apps, uses three or more apps from the catalog, or wants to avoid the ongoing cost of multiple individual subscriptions.

It probably doesn’t make sense if you run a stripped-down setup and already know exactly what tools you use. You’d be paying for optionality you won’t exercise.

For me, the answer is yes and the honest audit confirmed it rather than rattling it. I use enough of the catalog often enough that the subscription is earning its place.

If you want to run the same test, start with the apps you already use or would otherwise buy directly. If the list gets to three real apps, Setapp is worth trying. If it does not, skip it. The referral link is not a command to subscribe. It is a shortcut to run the same audit.

That’s the test worth running on any subscription: not “is this good?” but “is this earning its place?”

Disclosure: This article contains a referral link. If you subscribe through it, I may earn a small commission or account credit at no extra cost to you.

That’s the whole point of the metaphor. It takes what catches its eye from wherever it finds it… a bit of wire, a scrap of foil, something shiny someone else dropped and builds something new with the pieces.

But there’s a craft to stealing well. Grab everything indiscriminately and you end up with a pile of junk instead of a nest. Steal nothing and you’re starting from scratch every time, ignoring the accumulated work of everyone who figured something out before you.

In Steal What Works, Leave the Rest, I laid out the philosophy. This is the operational version: how to actually evaluate what’s worth stealing and what to leave behind.

The Problem with Productivity Advice

Most productivity systems come packaged as complete answers. GTD says: here is how you process everything. PARA says: here is how you organize everything. Zettelkasten says: here is how you connect everything. Atomic Habits says: here is how you build everything.

These are good systems. They’re built by smart people who thought deeply about how they work. They’re also designed for a hypothetical person whose life and workflow resemble the author’s. That person is probably not you.

The impulse to adopt a system wholesale is understandable, it’s much easier than doing the work of evaluating each piece. But wholesale adoption means you get the parts that fit your life and the parts that don’t, welded together in a package you feel obligated to maintain completely.

The alternative is selective theft. Take the parts that solve a real problem in your actual workflow and leave the rest.

What Makes Something Worth Stealing

A piece of a system is worth stealing if it meets three criteria:

It solves a problem you actually have. Not a problem the author had, not a problem you might have someday, a problem you’re experiencing right now. If you don’t have a capture problem, you don’t need GTD’s inbox processing ritual. If you’re not managing multiple active projects, you don’t need PARA’s full folder structure.

It’s smaller than the system it came from. The best steals are atomic. Not “I’m adopting GTD” but “I’m adopting the two-minute rule.” Not “I’m going full PARA” but “I’m going to keep a Projects folder and an Archive folder.” Atomic steals integrate without forcing you to adopt the surrounding context.

You can evaluate it within two weeks. If a habit, tool, or practice doesn’t show a measurable result in two weeks, you’re either implementing it wrong or it doesn’t fit your workflow. Good steals produce feedback quickly. Systems that require months of faithful implementation before you can evaluate them are selling you faith, not method.

What to Leave Behind

The flip side is knowing what not to take.

Don’t steal what requires context you don’t have. Zettelkasten works beautifully for researchers and writers who are building arguments across hundreds of linked notes. If you’re a project manager who needs to track deliverables, the Zettelkasten linking workflow will feel like overhead without payoff. That’s not a failure of Zettelkasten. It’s a mismatch.

Don’t steal what solves someone else’s problem. The person who built a complex inbox zero ritual was drowning in email. If email isn’t your pain point, their ritual isn’t your solution, no matter how elegant it is.

Don’t steal what you admire but won’t use. There’s a category of productivity practice that I find intellectually interesting and would never actually do. A daily full review of all projects and commitments. Color-coded task labels. Manual time logging. I’ve tried all of these and abandoned all of them. I still think they’re good ideas. They’re just not my ideas.

The Actual Practice

When you encounter a system, tool, or practice that interests you, run it through the three-question filter before adopting it:

1. Do I have the problem this solves?

2. Can I take a piece of it without taking all of it?

3. Can I evaluate it in two weeks?

If yes to all three: steal it. Try the piece. Evaluate honestly. Keep or discard.

If no to any of them: appreciate it from a distance and move on.

The nest you’re building is yours. It should look like your life, not someone else’s framework installed wholesale and maintained out of obligation.

Last week I sat down to count them. I knew the number was high, but I’d never actually added it up. Two hundred and fifty. That’s how many small web apps I’ve built and put online at southforkapps.com, free and no login required. A BMI calculator. A cron expression builder. A guitar fretboard. A thing that turns text into Pig Latin. Two hundred and fifty of them.

Here’s the part nobody tells you about building a lot of small tools: you don’t use most of them. I built a resistor color code reader one night because I was curious whether I could. I have never once needed it since. I wouldn’t could that as a failure as it’s the whole point, and I wrote about why in Disposable Software. Most of these were built to scratch one itch on one afternoon and then left alone.

But three of them stuck. Three I actually open on purpose, more than once, because they solve a problem I keep having. Those three are worth talking about, because they tell you which tools are worth building in the first place.

1. Recipe Scaler

I cook for six people. Four kids, my wife, me. Almost every recipe online is written for four servings, sometimes two, and the math to scale it up is the kind of thing I will get wrong while standing at the counter with raw chicken on my hands.

So I built Recipe Scaler. You paste in the ingredient list, tell it you want 1.5x or double or two-thirds, and it rewrites the quantities. That’s it. No account, no saving, no “premium tier for unlimited recipes.” I open it on my phone while the pan is heating up.

It survives because it removes a specific friction at a specific moment. The recurring annoyance (bad mental math under time pressure) meets a tool that’s faster to open than it is to do the arithmetic.

2. Teleprompter

Part of my job is making videos. When you’re reading a script to camera, the choice is usually between memorizing it (slow, and I’ll fumble it anyway) or glancing off to a sheet of paper (which looks exactly like what it is). Professional teleprompter software is either expensive or bloated with features I don’t need.

So I built a Teleprompter that does one thing: scrolls text at a speed I set, full screen, big enough to read from a few feet back. I paste the script, set the pace, and record. It runs in a browser tab on whatever machine is closest.

I use this one at work, which is the real test. A tool you only use in your own free time is a hobby. A tool you reach for when the clock is running and someone’s waiting on the footage is actually useful.

3. Pomodoro Timer

I’m forgetful and I’m easily pulled off course. Left alone, I’ll start writing a newsletter, remember I need to send an invoice, open a browser tab to do it, see an email, and forty minutes later I’m somewhere I didn’t mean to be.

The Pomodoro Timer is the dumbest of the three and the one I lean on most. Twenty-five minutes on, five off. The only thing it does that the timer on my phone doesn’t is sit in a tab and not tempt me to pick up the phone, which is where the trouble starts. I run it during writing blocks and during the boring stretches of my second job, where having a clock counting down keeps me moving instead of drifting.

It’s not sophisticated. It doesn’t track my focus history or sync to a dashboard. I don’t want it to. The moment a focus timer needs me to focus on it, it has failed.

What the three have in common

I didn’t pick these on purpose. They earned their place by being the ones I kept coming back to. Looking at them together, the pattern is obvious:

• They solve a problem I have on a schedule, not once. Cooking, recording, focusing. Things that recur.

• They open faster than the problem takes to solve manually. Zero friction. No login, no setup.

• They do one thing. None of them try to be a platform. The recipe scaler will never have a social feed.

That’s the lesson hiding in 247 tools I don’t use. Building software got cheap enough that I can make a thing in an afternoon to solve a single annoyance, and most of the time that’s all it ever needs to do. The 247 weren’t wasted. They were the cost of finding the 3.

If you’ve got a recurring annoyance and a free afternoon, build the dumb little tool. Worst case, it joins the pile. Best case, it’s one of your three.

All 250 are free at southforkapps.com. Go steal the ones that solve a problem you actually have or use what I have already built for free.

I’ve been running an AI agent against my Obsidian vault for a few months. Not a chatbot that answers questions about my notes. An agent that opens files, reads them, rewrites them, and creates new ones. It runs on a schedule. It does this while I’m at work, sitting in front of the computer, or asleep.

This is not theoretical. This is my actual setup. And I want to walk through what it actually means before the think pieces about “agentic AI” turn it into something unrecognizable.

What “agentic” actually means

When most people talk about AI and their PKM, they mean asking ChatGPT to summarize a note they paste in. That’s useful. It’s also just a slightly faster search bar.

An agent is different. It has access to tools and in this case, the ability to read files, write files, run shell commands, and call external APIs. When I talk to my agent, it doesn’t just process what I paste. It can go find the thing. It can check yesterday’s journal entry, read three related files, and write a summary without me feeding it anything.

The technical setup is Claude Code (Anthropic’s CLI tool) plus Model Context Protocol, which gives the agent the ability to interact with external systems. For Obsidian, that means it treats the vault like a filesystem because that’s basically what an Obsidian vault is. If I don’t want to use a robot to edit a note, I can just go in myself to create or edit a note.

What it actually does

Let me be specific, because “AI manages my notes” sounds either amazing or horrifying depending on how you hear it.

Here’s what my agent does on a regular basis:

Generates content ideas automatically. Every Monday morning, it searches productivity communities and recent articles, pulls out ten content ideas for this newsletter, and writes a dated file into my vault. I find it there when I open Obsidian. I didn’t ask for it that day. It just happened. (This very article started as one of those automatically generated ideas.)

Transcribes and processes podcasts. My podcast workflow is fully automated. New episodes are downloaded, transcribed locally, and written into my vault with show notes and a full transcript. I open a note and the work is done. I have links to everything the hosts talked about whether they linked them in the show notes or not.

Maintains a session log. After every work session, the agent writes a summary of what happened, what was done, what changed, what’s pending. The vault has a running memory I don’t have to maintain myself. There is no way that I could keep up with everything that it’s creating. So, I let it maintain itself.

Acts on inline instructions. If I write @COG: anywhere in a note, it picks that up and executes the instruction the next time I interact with it. For example, whenever I am dropping a note in my inbox about making a meal plan, I can put @COG: research health meals for four kids and put them in a list for the week so that the kid’s meal plan is done and then I can add whatever I want on the menu and the AI does nothing to what I write. I don’t have to remember to follow up. I flag it and move on. This also helps out determining whether I just had an idea that I need to work on or if it’s something that I want the robot to just do without having to ask me.

None of this required me to write code. I set up the system in plain markdown files, basically a set of instructions that tell the agent what to do and when.

The part that actually matters

The thing people get wrong about agentic systems is treating it as an all-or-nothing question. Either AI does everything or you don’t let it touch anything. The practical reality is that you hand it tasks with clear, bounded success criteria and you keep it away from decisions that require your judgment. It’s going to make mistakes. It’s going to misinterpret a request. That’s why you need to review what it makes and ensure that the output is what you wanted.

My agent can write a draft. I edit it before it goes anywhere. It can create a summary. I decide whether it’s accurate. It can flag a note for review. I decide what to do with it. What I don’t let it do: delete anything permanently, send anything external, or make judgment calls about what matters. Those stay with me. The guardrails are not complicated. They’re just decisions you make once about what “done” looks like for a given task.

Whether this is for you right now

I want to be honest about the current state: this setup requires comfort with a terminal, some tolerance for friction during configuration, and a willingness to think in systems rather than apps.

If you’re not there yet, that’s fine. What I’d suggest is watching this space closely, because the friction is dropping fast. Also, I’m not the only person doing this already. Six months ago this was significantly harder to set up than it is now. In another six months it will probably be a few clicks.

What you can do now, regardless of technical level, is start thinking about your vault differently. Not as a container you fill and occasionally search, but as something that could participate in your work. What would you want it to do automatically? What decisions would you never hand off?

Pick one task in your vault you do manually every week. Write down exactly what done looks like. That’s the first thing worth automating.

That sounds obvious, but it explains why so much of it falls apart the second real life shows up. The advice assumes you can sit down for a weekly review with coffee and a clear desk. It assumes you have an hour to rebuild your task manager. It assumes your problem is that you have not found the right framework yet.

Sometimes the real problem is that you have eleven minutes before the next thing, your brain is tired, and every task on your list looks bigger than the amount of energy you have left.

That is the situation the 5-Minute Task System is built for.

The Trap of the Big System

I love productivity systems. I wrote about my larger setup in Building a Magpie’s Nest: My Scavenged Productivity System, but this is the smaller emergency tool inside that bigger approach. I would not have started The Productive Magpie if I did not enjoy picking through tools, templates, apps, workflows, automations, and weird little methods that make life run better.

But there is a trap in all of it.The bigger the system gets, the harder it is to use when you are already behind. When your task manager has twelve areas, four priority levels, custom filters, nested projects, contexts, tags, energy states, someday lists, and recurring reviews, it can be wonderful on a calm day. It can also become one more room you have to clean before you are allowed to do the work.

That is backwards.

A productivity system should help when your day is messy. If it only works when your life is quiet, it is not a system. It is decor. The 5-Minute Task System starts from a different assumption: you are busy, interrupted, and probably carrying more open loops than you can comfortably name. So instead of asking, “What is the perfect way to organize everything?” It asks, “What can I move forward in five minutes?”

Why Five Minutes Works

Five minutes is too short to intimidate you.

It is also long enough to matter.

You can answer the text.
You can put the receipt where it belongs.
You can unload part of the dishwasher.
You can write the ugly first paragraph.
You can open the document and leave yourself the next sentence.
You can schedule the appointment.
You can pay the bill.
You can rename the file.
You can send the follow-up.

None of those feel like a dramatic productivity breakthrough. That is the point.

I had to call about my medications to answer questions as required by my insurance. I had been putting this off all day because it meant talking on the phone with another human being. As I was leaving the studio, I just decided I was going to get it done, so I pulled out my phone and was finished with the call by the time I got to my truck. Something that had been bothering me all day was completed in under 50 steps.

Most of the pressure in a busy life does not come from one giant task. It comes from the pileup of tiny unresolved things that keep asking for attention in the background.

The unanswered message.
The thing you need to return.
The form you need to fill out.
The appointment you need to make.
The note you need to capture before you forget why it mattered.

Each one is small enough to dismiss. Together, they turn into static. The 5-Minute Task System is a way to lower the static.

The System

Here is the basic version. Make a list of tasks that can be moved forward in five minutes or less. Not completed necessarily. Moved forward. That distinction matters.

“Clean the garage” is not a 5-minute task. “Throw away the cardboard by the garage door” is.

“Write the newsletter” is not a 5-minute task. “Write three possible openings” is.

“Get my finances together” is not a 5-minute task. “Download the bank statement” is.

The move is to break tasks down until the next action is so small that your brain stops arguing with it. Then, when you have a weird gap in the day, you do one.

Waiting on coffee? Do one.

Five minutes before a meeting? Do one.

Sitting in the car before walking inside? Do one.

Your energy is low and you cannot face the big thing? Do one.

The Magpie Version

The Magpie Method is not about pledging loyalty to one productivity religion. It is about stealing what works and leaving the rest.

This system steals from a few places:

• Getting Things Done gives us the next action.

• Atomic Habits gives us the idea that small actions compound.

• Bullet Journaling gives us permission to keep the list simple.

• Plain old survival gives us the truth that busy people need systems that work in bad conditions.

I wrote more about that identity side in The Identity of a Productive Magpie, but this system is the ground-level version: one small action, then another. I do not care whether the list lives in Obsidian, Apple Notes, Todoist, TickTick, Notion, a sticky note, or the back of an envelope. Use whatever you will actually open.

What Goes on the List

A good 5-minute task should be specific enough that you know exactly what to do when you see it.

Bad:

• Work on taxes

• Clean kitchen

• Fix website

• Plan content

Better:

• Find W-2 and put it in tax folder

• Clear the left side of the counter

• Check if the homepage button works

• List five newsletter headlines

The test is simple: if you read the task while tired and still understand the next move, it belongs on the list.

If you have to think, rewrite it.

What Does Not Go on the List

Do not put fake 5-minute tasks on the list. You know the kind.

“Outline entire course.”

“Organize all photos.”

“Fix my life.”

Or one that I find myself using often, “Admin tasks.”

Those are not tasks. The system only works if you are honest about size. Five minutes means five minutes. Maybe seven if you are already moving. If it turns into forty-five minutes, it was not a 5-minute task. Break it smaller and tackle the smaller parts so they add up.

Why This Helps Busy People

Busy people do not always need more ambition. They need fewer stalled objects in the path.

When you are juggling work, family, side projects, appointments, errands, health stuff, house stuff, and the strange administrative fog of modern life, your brain gets tired of holding all the tiny things.

The 5-Minute Task System gives those things somewhere to go. More importantly, it gives you a way back in. That might be the most useful part. When you fall off your system, you do not need a grand reset. You do not need to spend Saturday rebuilding your Notion dashboard. You do not need to declare a new era. You need one small task you can finish. Then another... Then another. Momentum usually returns after movement, not before it.

A Simple Starter List

If you want to try this today, start with ten.

• Reply to one message

• Put one loose paper where it belongs

• Clear one surface

• Start one load of laundry

• Write one bad paragraph

• Rename one file

• Schedule one appointment

• Pay one bill

• Capture one idea

• Delete or archive ten emails

Do not make this precious. Do not build the perfect template first. Do not go app shopping. Write ten small tasks and do one of them. This is the same reason I love tiny systems like my Q shortcut. Small bits of saved friction add up.

The Product

I turned this into a small product because I wanted the system to be easier to start than to overthink.

The 5-Minute Task System for Busy People gives you a ready-made structure for capturing tiny tasks, breaking down bigger ones, and keeping a short list of actions you can use when the day gives you a spare pocket of time.

It is not a life operating system.

It is a pressure valve.

You can grab it here:

The 5-Minute Task System for Busy People

If you are already behind, do not start by rebuilding everything.

Start with five minutes.

Disclosure: This article includes an affiliate link. If you sign up through it, I may earn a commission at no extra cost to you.

If the button above does not work click here: The 5-Minute Task System for Busy People

One of the reasons that I love the letter Q is because it is very easy to find on the keyboard. Q sits right next to my left pinky. I could find it blindfolded.

I work with my left hand on the keyboard and my right hand moving back and forth from the keyboard to the mouse. My left hand is always great for text automation macros and quick access because of this.

Why Q Works

It’s the best letter for automations and macros because it is one of the least used letters in the alphabet. When I started setting up shortcuts, I used semi-colon, period or Z. These all cause conflicts at some point with words or with other shortcuts like semi-colon being used for emojis in Slack.

You may be asking why the letter Q when X and Z are right there too! My thought is that Q is almost always followed up by the letter U, which no other letter follows as much as the letter Q does. Doubling up on Q (qq) makes it even rarer as no English word starts with that.

How I Use It

You might be curious about how a letter can be used as an automation, but with Keyboard Maestro, you can set up hot key automations based on text strings. I can type the letters ‘qq’ and then start typing words that can trigger automations. I have almost 100 macros set up that start with ‘qq’. It’s as simple as typing ‘qqaddress’ which automatically inserts my address.

Need some examples?

• qqemail → your email

• qqsig → email signature

• qqbio → short bio

• qqcal → “Here’s my calendar link: ...”

You don’t use Keyboard Maestro? No problem, simply add it to Apple’s text replacements and you have access on all your Apple devices with a quick and easy text replacement to save time and your thoughts.

Setting It Up

How to set up text replacements on Mac:

1. Open System Settings → Keyboard → Text Replacement

2. Click the “+” button to create a new replacement

3. In the “Replace” field, enter your shortcut (e.g., qqphone)

4. In the “With” field, enter what you want inserted (e.g., your phone number)

5. Click “Done” and the replacement is live immediately

How to set up text replacements on iPhone/iPad:

1. Open Settings → General → Keyboard → Text Replacement

2. Tap the “+” in the top-right corner

3. In the “Phrase” field, enter what you want inserted

4. In the “Shortcut” field, enter your trigger (e.g., qqphone)

5. Tap “Save” and it syncs via iCloud to all your devices

The beauty of this approach is that it works everywhere! Mail, Notes, Messages, browsers, any app with a text field. Once you set it up, you never have to type those long strings again. It saves me from typing long strings and from forgetting the ones I’d otherwise have to look up.

Quick Recall Markers

It’s also great at finding sections of written work, so if I need to come back and review something, I can type QRM for Quick Recall Marker. I got this inspiration from using TK because the letters T and K are rarely put together in any English word, but I’ve but some people I know have names that contain TK, which tripped me up. Honestly, I would forget the letters also. TK is a journalism/editing convention which means “to come” as a placeholder for missing info. This got me thinking of what I could use.

So, I use QRM in any text field to quickly recall something or bring my attention to the surrounding text. QRM only shows up as technical acronyms and never as informal English words. This makes it a great way to search and find text that I might want to recall later. QRM sticks in my head because it actually means something and it’s not just random letters to me.

I use this whenever I need to come back and check something that I’ve written, or as a way to remember what I’ve done. For instance, if I’m writing a story and need to fact check, like I did earlier on the letter Q being the least used letter in the alphabet, I just type QRM and come back to check later. Whenever I’m ready to post, I search for QRM and jump right to it.

The Real Win

Small systems like this (finding the right letter, setting up one shortcut) compound into hours saved over a year. The real win isn’t Q or QRM. It’s choosing a system that fits how you work and sticking with it.

Be a Magpie and steal the letter Q from me.

It also knows not to use em dashes. I hate em dashes.

This didn’t happen by accident. I built it that way. And today I’m packaging the whole thing so you can use it too.

The Problem With AI Assistants

Every AI tool you’ve used has the same flaw: it forgets you the moment the conversation ends.

You spend five minutes giving it context. It gives you a great answer. You close the window. Next session, you’re a stranger again.

The productivity community’s solution to this has been prompt templates and custom instructions. Paste your context at the top of every conversation. It works, barely. It’s also exhausting and still doesn’t account for anything that changes day to day.

I wanted something different. I wanted an assistant that actually knows me and gets smarter about me over time without me having to repeat myself constantly.

Obsidian as the Brain

I already lived in Obsidian. My notes, my journals, my health log, my media library. All of it was there. The problem was it was static. I could find things, but nothing was connected to action.

The insight was simple: what if I gave Claude a folder in my vault and told it to treat that as its memory?

That’s the whole idea. Claude doesn’t “remember” me but it reads files that do. Every session it boots up and reads:

• A file about who I am (family, job, health conditions, preferences, communication style)

• A running log of every session we’ve ever had

• A list of active projects and where they stand

• A set of skill files that tell it exactly how to handle specific tasks

That last part is the key. The skills aren’t prompts. They’re documented workflows that are step-by-step instructions for recurring tasks, stored as plain markdown files in the vault. There’s one for morning briefings. One for logging health data. One for processing my inbox. One for logging movies and TV. One for weekly reviews. One for shutdowns.

When I say “log a movie,” Claude reads the media-log skill, pulls metadata from the web, creates a properly formatted note in my /Movies folder, and moves on. No fumbling, no clarifying questions, no “as an AI language model.” It just does it.

What This Looks Like in Practice

My mornings start the same way. I open Cowork (Claude’s desktop app), type “good morning,” and within about thirty seconds I have:

• Current weather and what’s coming later in the day

• Everything on my calendar pulled from all my Google calendars in parallel

• A quote worth reading

• Any open loops from the last session I need to know about

• An inbox count

Then it asks if I watched anything last night that needs logging. If I say yes, it handles the whole thing (title, cast, gene, cover image, rating, my notes) and files it in the right place.

Why This Works When Other Setups Don’t

I’ve tried the custom instructions approach. I’ve tried pasting context blocks. I’ve tried dedicated AI apps that promise memory features.

The difference here is that the memory lives in a place I already control, in a format I already use, and it’s updated automatically as part of the workflow itself or I can go into the .md file and edit it myself.

I don’t have to use tokens if I want to make a change.

When I log something, the log gets updated. When a project changes status, projects.md gets updated. When I shut down, a session summary gets written. The system maintains itself because maintaining it is built into every skill.

It also means I can read my own logs. I can see what Cog (that’s what I call my robot) did two weeks ago. I can search my vault for any decision I made or any task that came up. It’s not locked in some proprietary memory system I can’t inspect or export.

The Magpie Vault

I’ve been running this system for a while now. It’s become the thing I use more than anything else on my computer.

So I packaged it all up.

The Magpie Vault is an Obsidian starter pack built around this exact setup. It includes:

• The full Robot Assistant skeleton CLAUDE.md, all the core skill files (morning, journal, health, media log, weekly review, inbox, shutdown, and more), and the memory file structure

• A health tracker template pre-built for daily logging

• A media log template for movies and TV

• Weekly and monthly review templates

• A full README walking you through setup step by step

You bring your own Claude account. Everything else is in the vault.

It’s $19. One time. You own it.

Get the Magpie Vault →

If you’ve been wanting to use AI more intentionally and not just for one-off questions but as an actual part of how you work, this is the place to start. The skeleton is already built. You just have to fill in the details about yourself and let it run.

That’s the Magpie Method. Steal what works, leave the rest. This one works.

Questions? Hit reply. I read everything.

When the doctor told us she was allowed 4,000mg of acetaminophen every 24 hours and not to exceed that, I went and asked it to keep track of how much acetaminophen was in each of her medicines. I was able to put that in, and then it gave me a graph at the top showing how much she had taken in the past rolling 24 hours and how much she was allowed to take.

Once things calmed down, I was able to look around on the App Store to see if there was something on the market that gave me this level of customization and I found nothing. There were no apps that met everything I needed, yet this application that took me 30 minutes to get started and 20 minutes here and there to customize was exactly what I needed when I needed it. Her surgery went well and pain management was phenomenal, so she was able to get off the pain medicines within a week. I just didn’t open the web app anymore.

I didn’t have to uninstall anything, didn’t have to unsubscribe, didn’t have to pay for any of it (other than my Claude subscription). Had I downloaded an app, there more than likely would’ve been a subscription, a one-time fee, or some kind of payment for something I only used for one week.

I don’t know what that means for app development and what it means for the future of software, but I was blown away by how easy it was to get it going, customize it to exactly what I needed, and use it for when I needed it.

Before this, my default move was opening the App Store and searching until I found something that was close enough to what I needed (and then paying for it anyway). Now my default move is opening Claude and just building it. I wonder how many of you are still searching the App Store when you could just build exactly what you want in 30 minutes. Are you still doing it the old way?

I am still in the early stages of the Productive Magpie concept. I have learned not to hoard tools but to build systems that reinforce who I am and how I work. After finishing Atomic Habits by James Clear, this concept was further fleshed out as I looked to build in those systems into my magpie approach.

Identity Over Goals

One of the ways that has helped me is to think that “I am a writer” instead of “I want to write blog posts.” I could create a goal to write this blog post, but once I have finished it, I would have to set a new goal of another blog post. Constantly moving the goal posts is not beneficial. So, I think of my system. I write my ideas in Ulysses and then start writing what is on my mind. I might not finish the post, but I have the idea out of my mind and into something functional. You need to move away from goal setting and work toward identity building.

Downloading TickTick doesn’t make me organized,

but using it every day does.

A magpie doesn’t collect things because it wants the most things, it collects things to build a stronger nest. In my previous blogs, I would focus on one blog post and set that as a goal to write one blog post at a time. I thought that was being productive. I now realize that I was not being a writer, I was just writing. That may be why all those other posts failed and why I quit writing there. Writing was a chore, it wasn’t just what I do.

Clear famously says, “You do not rise to the level of your goals. You fall to the level of your systems.”

The Magpie Workflow

You have to create a system to get to where you want to be. I have an input system of RSS feeds that bring in articles of interest for me to read. I weed out what I want to read by sending the articles to TickTick so that I can make any notes in the description as well as setting a due date if it is something that I want to focus on. This also allows for articles to sit until I have time (like on a weekend morning) to get to them. Sometimes these gems of articles and ideas pop up then that didn’t resonate when I skimmed them throughout the week. Once I have an idea, I can send that topic over to Ulysses and start typing away.

The Power of 1%

James Clear also emphasizes that if you improve by just 1% every day, you will be exponentially better within a year’s time. That is my hope for the Magpie method. I want it to improve and for others to discover the benefits of stealing from different areas and actions that they come across. So, if I learn one keyboard shortcut today, that will help me work faster the rest of the year possibly saving me hours throughout the year. I could take the time to better my system by creating a template so that later on when I need to work on a similar project, I can implement that template and cut out some of the duplicate work that I would have to do if I built it from scratch. Those 1% improvements add up. That is the essence of being a productive magpie, gathering small wins that add up to a massive advantage.

What is one 1% improvement you’re making to your system this week?

Watch a magpie build its nest and you’ll see a creature that refuses to be loyal. Magpies are thieves with taste. They don’t commit to one tree or one yard. Instead, they scavenge, hopping from fence post to garden to parking lot, selective about what catches their eye. A shiny wrapper from the parking lot, a bit of wire from the construction site, a lost button from the playground. Each scavenged bit comes from somewhere different, and none of it matches.

Together, it builds something functional: a nest that works.

The magpie’s superpower isn’t finding one perfect source of materials. It’s knowing what to steal and what to leave behind. It takes what serves its purpose and moves on, building a nest from a dozen different places.

This is exactly how I approach productivity. Not committing to one system, but stealing the best pieces from everywhere. Being productive means different things to different people. Most productivity advice pushes you toward one ecosystem: Apple or Google, Notion or Obsidian. That’s the trap.

Breaking Free from the Algorithm

I use the app Reeder as my only view into social media and the news. It’s a finite feed, and once I read through it all, I’m finished. There’s no endlessly refreshing and pulling up new content that the algorithm thinks I might like. I read from people I want to follow and sources I want to learn from. This keeps me from wasting time endlessly scrolling through the news or slop of the day.

I’ve spent years on social media, RSS, and simply reading articles on the web. I’ve found sources that have come and gone. In doing so, I’ve resonated with certain writers, so I follow those writers wherever they contribute. Unfortunately, most still use some type of social media that doesn’t integrate with Reeder (looking at you, Instagram), which makes it difficult or impossible to see everything they post. I’ve learned that’s okay. I don’t have to see everything all my favorite people put online. If it’s something they find important, they’ll post multiple times or mention it somewhere else so I pick up on it. And if not, I simply don’t get that content. The world keeps spinning.

When I first switched from an algorithmic feed to a finite one, the biggest change I noticed was how much time I gained in my day. I had set up lists on Twitter to only see people I wanted to, but what did the app do when you first opened it? Go straight to the algorithmic timeline. And of course, something interesting was at the top of the list, and away I went scrolling down to get another hit of dopamine.

I borrowed the best of online social connection while rejecting the algorithmic trap.

Scavenging Software That Works

As I stated in a previous post, I’m using Setapp. This post contains paid links. If you purchase Setapp through my link, I receive a commission at no extra cost to you. This allows me to test many different applications to find what suits me best for one subscription. The Magpie approach to software means I don’t have to commit to one ecosystem. Instead, I use Setapp to find different applications that serve different aspects of my life.

I write my posts in Ulysses, which is part of Setapp, because I like the way I can have material articles marked as references for my posts. The design is minimalist while having Markdown support. I brought this into my workflow after testing Obsidian, Notion, and others. I learned that I don’t like block-based text editing and love writing in Markdown.

Another app that’s part of Setapp is Craft. I use this tool to keep up with work projects. I like it because I’m able to create a Notion-style database within a document. This allows me to keep my tasks together and see visually where the project currently is. I’ll dive deeper into these tools in the future.

Your Turn

Productivity isn’t about finding one perfect system that solves everything. It’s about being a Magpie: picking the best bits from various sources and building a workflow that actually serves your life instead of consuming it. Take a look at your own digital diet this week. Are you scrolling through an endless algorithm, or are you intentionally choosing what to consume so you can get back to the things and the people that actually matter?

Build a nest from diverse sources instead of living in someone else’s pre-fabricated cage.