One app category down

A personal AI with a todo skill just replaced 1000+ todo apps. Here's the pattern.

Watch me build it step-by-step on YouTube

There are over 1000 todo apps in the app store right now. Todoist alone has 40 million users.

And yet most people still forget to do things.

I analyzed a lot of reviews for these apps. Here's what people actually complain about:

Everything is still manual. You enter everything. You prioritize everything. The app just holds your list.

No intelligence. Keep postponing a task? The app doesn't adapt. It just nags.

Overwhelming lists. 47 overdue items. You feel anxious. You close the app.

Siloed data. Your todos don't talk to your calendar. Your calendar doesn't talk to your email. Everything is scattered.

People forget to open it. This is the most common complaint. The app only works if you use it.

These problems aren't bugs. They're architectural. Todo apps are containers for lists. That's all they can ever be.

What Replacement Looks Like

I just built a personal AI with a todo skill using n8n and Arca. Took about 10 minutes. Just taught my AI how to manage todos by giving it a skill and access to a data vault.

I recorded the whole thing so you can build your own personal AI the same way: watch the tutorial

Now I say things like:

• "What's due this week?"
• "Add a todo: Call mom on Sunday"
• "Mark the quarterly report as done"
• "What's my highest priority task?"

My AI reads the skill, queries my data, and responds. No app to open. No UI to learn. Data lives in my private vault. Exportable anytime.

That's my setup. But the pattern is what matters.

The Pattern

Every app category follows the same structure:

1. Data — What gets stored (tasks, due dates, priorities)
2. Logic — How to interact with that data (add, complete, query)
3. Interface — How the user accesses it (app UI)

Traditional apps bundle all three. That's why they're silos.

Personal AI unbundles them:

1. Data — Lives in your private vault (Arca)
2. Logic — Your AI reasons about it using skills
3. Interface — Whatever you're already using (chat, voice, SMS)

The skill is just a small file that tells your AI what this data is and how to help you with it. When you ask "what's due today?" your AI reads the skill, queries the table, answers.

One skill replaces one app category. Maybe several.

Why This Matters

Todo apps are a $2B+ market. But they're also the simplest possible example.

We're building skills for:

• Meal tracking and calorie logging
• Weight logs
• Workout tracking
• Journal entries
• Grocery lists
• Personal CRM
• Family management
• Travel planning
• Any type of life logging
• ...many more

Each one follows the same pattern. Data table plus skill file. Works with any AI that supports MCP—Claude, ChatGPT, custom agents, whatever comes next.

The apps in these categories won't disappear overnight. But they're going to feel increasingly unnecessary. Why download an app when you can just tell your AI to do it?

The Shift

When I wrote "Apps No More" a few weeks ago, some people thought it was too aggressive. Apps are a $500B market. They're not going anywhere.

I think they're looking at the wrong thing.

Apps won't die because they stop working. They'll die because people stop needing them. The same way we stopped needing standalone GPS devices after Google Maps. The same way we stopped needing portable music players after smartphones.

The container changes. The data moves.

What we're building is the infrastructure for that move. A place for your data to live that isn't locked inside any single app or company. A way to teach your AI new skills without writing code. A pattern that works across every category.

One app category down. A few hundred to go.

Feel free to reply to this email or DM me on LinkedIn to jam on these ideas.

If you want to see how this works, check arca.build. Or if you'd rather skip the building and just use a personal AI that already has this built in, there's mio.fyi.