DEV Community: Sagi h

Personal software - solved

Sagi h — Fri, 30 Oct 2026 09:31:57 +0000

For decades building software was reserved for a select few - those who dedicated their lives to the craft, who enjoyed it, who practiced it. Not anymore.

A new era of software is upon us.
Some deny it and some see it as clear as day.
Some are thrilled by it and some dread it.
It will affect everyone for better or worse, but mostly better.

You know the feeling when you use something you yourself built, as opposed to when you use something that was handed to you. You feel in control, you feel safe. You know exactly what to do, when and how, and you can easily navigate it. You cannot hate it, because if you dislike something - you just fix it.

The feeling is unparalleled, but for the longest time software was out of reach for most. AI closed the coding gap, so anyone can tell it what to build - but still, something was missing. Every piece of software comes with one:

The Tail

A project with dozens of files just to run a simple text box
Servers
Security
Hosting
Compiling
Launch sequence

It's easier than before for sure, but it's still a lot of preparation for the simplest app.

Where is that damn executable file? What do you double-click on?

There is only one correct way to use an app: double-click it.
And then it hit me - we can cut the tail for personal apps and keep the double-click. We can have our cake and eat it too.

The tech is already built but it is scattered.
We have all the ingredients but no one ever combined them.

Until now.

The PaPP Era

Apargon - open-source, completely free, and AI-first software.

Apargon is a new software primitive, a pillar, and it unlocks a whole new industry of apps. In technical terms it is a "Sandboxed GUI", meaning a locked window. Nothing can come in and nothing can come out unless you approve.

Apargon removes the tail. No more hosting, one executable file, no servers, and the launch sequence is a double-click. It can't get simpler than that.

Now you can download Apargon and tell your AI agent to build the app you've always needed. A personal app (PaPP) is an app that solves the exact problem you face, not something someone else prescribed for you, but your ideal process, your ideal solution.

All you need is a problem to solve. AI codes, Apargon runs.

Understand this. We already have browsers like Chrome and Edge; those who know how can create HTML files and double-click them. But there's one fatal flaw: browsers do not let a page access your computer freely. Either you run a server, or you keep going through your Downloads folder.

Apargon solves it with permissions. Apargon denies everything by default, and you explicitly allow what the PaPP can access - or rather, what your agent asks for. All you do is confirm a popup, just like when you sign in with Google. The popup tells you what the PaPP requests.

You ask your agent to build a PaPP, and that's it.
One .papp file, one app. Double-click and it opens.

There are endless possibilities. The only limit is your imagination.
All of your problems, challenges and issues are now one or two prompts away.

Want a budgeting app without the subscription?
Want to visualize your health data?
Want to see your run drawn on a map?
Want a quick video editor, or an audio trimmer?
Want to create a command center for your smart home?
Want an email client for your domains?

Just tell AI what you want it to build for you, and watch it cook.

SaaF

We are entering what is called SaaF - Software as a Fashion.

China revolutionized textiles and clothing by making all clothes cheap, fast to produce and in high volume - fast fashion. AI and Apargon will do the same for software. Built a PaPP? Done using it? It broke? No hard feelings, throw it away and ask AI for a new one.

High-end designers will remain for high-stakes software like health and finance, just like in fashion, where they exist for very specific types of clothes. But we, the people, will be able to get the shirt and pants we need for the right weather at the right time.

Build your own

I beg you. Do not overthink it, do not doubt yourself, open your mind, and see that all of the current apps that are out there do not fit you well. Every single one of them has a flaw that you need to bypass or adjust to somehow.

Download Apargon, set up an AI agent and try for yourself. It cannot get easier than that.

The browser solved distribution. This was the easiest way to access the app, all you need to do is write the url and hit enter - easy. Desktop apps faded out of popularity for exactly this reason: where do you find the right app?

Even if you found one you still needed to go to the store or to the web site and download it - heck if you are already on the site, why isn't it a web-app?

Well, this is no longer an issue, because you do not need to find anything. With PaPPs you just ask AI to build it for you, and let Apargon run it for you.

Don't trust. Verify. Apargon is completely free and open source. Aim your best model at it, and if you find a bug, open an issue.

The last human skill standing

Sagi h — Mon, 26 Oct 2026 13:30:20 +0000

With AI anyone can build anything today. My 16-year-old cousin built her personal app with a little push in the right direction. Yet there is one skill that will stay human forever: Problem solving.

AI can build a solution, but for what? To solve a problem you must first see it.

There are three kinds of people:
People who do not know they have a problem.
People who learned to live with the problem.
People who see the problem and act to solve it.
The third kind is rare. Maybe you know someone who acted and fixed the thing everyone complained about.

But you? Even if you notice a problem, you wait for someone to hand you a solution on a silver platter. Maybe you look for an app on the app store or wait for the IT department to do something.

Until something is presented to you, you keep living in the discomfort and accept the situation. Even when something is presented to you, you still feel an itch - "Something is still off."

As with anything in life, you either adjust to the process, or the process adjusts to you. Why didn't you build the solution yourself? The app that would've solved it.

I'll answer that: you thought it was impossible, and you still do. You have a hard glass ceiling above you.

Honestly, it's not your fault: we all start there.

You were probably never taught how to approach a problem. Heck, you barely even know how to spot a problem to begin with.

As a software engineer, with over 10 years of experience and more solved problems than I can count, let me open your eyes and make you see your daily challenges, and how personal apps can save you time and money.

I'll share with you the exact thinking process of entrepreneurs, good software engineers and problem solvers to find a solution to a real challenge.

Know this - the skill can be learned. You can become the one who acts. Allow me to give you the same push I gave my cousin.

Good software solves a real problem

Look at your phone or laptop and see that it is full of apps someone else built. Each app tries to solve a particular problem you once had or still face, and the only reason you have them is that you have no choice. You adjust your process around the app. I bet there is more than one app you wish you did not have to use.

First you need to understand what types of problems there are, and what is considered "good software".

Good software solves one real problem, and there are three kinds of problems: core, specialized and what I call hyper-specialized problems. Core problems lie at the base of the others. Everyone faces those problems one way or another. E.g. communication - email, boredom - games, presentation - PowerPoint, data management - Excel.

Specialized problems are problems that not everyone faces, but many do. They are a combination of several core problems, for example:
Apps for money management, calorie tracking, the apps you use at your job.

Hyper-specialized problems are unique. Have you ever told yourself, "Damn, I wish this app existed - it would have saved me a ton of work", or "This app is so bad, there are 10 ways to do it better" or maybe you noticed a task that you do again and again that an app can automate?

There is a reason why there are no solutions for these problems. There is no money in them. No one faces them but you.

But this does not mean there is no value in solving them. The exact opposite, in fact. Here is where "personal software" comes into play. For a stranger, personal software that solves your hyper-specialized problem might be useless indeed, but for you, it is invaluable.

Personal software can save you a ton of time, completely automate a task you do, show you things you never knew about yourself. Honestly, the only limit is your imagination. Personal software can completely transform your life.

To solve a problem, you must see it first

We humans tend to confuse familiarity with understanding. Try it yourself - take zippers for example, since we are all familiar with them. First, rate yourself from 1 to 10, where 1 means you have no idea how it works and 10 means you make zippers in your backyard (if you actually make them in your yard, pick toilets or locks). Then explain it, step by step, and rate yourself again.

You quickly realize that there are holes in your understanding, and suddenly the clarity you felt disappears and the gaps appear. Leonid Rozenblit and Frank Keil ran this exact experiment in 2002. They named the phenomenon the "illusion of explanatory depth".

You might think you understand your morning, your job, or your evening. Write it down, step by step, and include the challenges and the problems you face.
What do you do repeatedly but don't like?
What is that workaround that you are proud of?
What's harder than it should be?
What is it you try to achieve?
What annoys you?

What is it you do in your job? Be specific, write the process down. For example:
A nurse who needs to triage patients in the ICU.
A lawyer who needs to be ready for a case.
An engineer who needs to solve something.
A shop owner who needs to keep track of his inventory.

The first step towards solving the problem is to stop reacting to life for a minute, and become a 10. You need to externalize the process - write it down, and see the problem reveal itself. Looking at the process is far more effective than thinking about it.

Now look at what you wrote with the simple assumption that nothing is good enough and everything can be better.

Pick one challenge that bugs you, and elaborate on the process you currently follow to achieve the goal. What are the apps you use that help you achieve it? How do you use them? Write down the exact process from requirements to achieving the goal.

Let's take Joe the copywriter as an example.

Joe's boss sends him an email with a task: "Write me a piece about how our new product is the best in the world."

Joe's favorite tool to write his drafts is his pen and paper. That's how he thinks best.

He drafts a 2,000-word article on how the new product is the best in the world, feels good about the draft, and needs to start editing.

Joe edits his work on his laptop, and for this, his tool of choice is his text editor - "Obsidian". He does not really like it, but it does the job. But there is a problem - he can't edit paper.

To get the text from paper into the software, he takes photos of his pages, then uploads them to Google Drive so he can download them to his laptop.

Once the images are there, Joe asks AI to transcribe each image to text, and then, after each transcription is ready, he copies each transcription, and creates a file for each in Obsidian to start editing.

You see, Joe adapts his process to the software. Every tool is a good solution to a particular problem Joe faces. But this entire process gets on his nerves, and takes like 30 minutes, from draft to edit - each time.

Good solutions on their own, but the process is still faulty - the time and effort are the problem.

The second step is to break the story into small actions, from the very beginning (the email) until the task (the copywriting) is complete. Write it in painstaking detail.

Joe gets a task from his boss via email.
Drafts with his pen and paper.
Takes photos of his pages.
Uploads them to Google Drive.
Downloads the images to his laptop.
Asks AI to transcribe each image to text.
Creates Obsidian files for each transcription.
Edits the files.

Once the story is written down, the third step is to create the magic chain.

The magic chain

A solution to a problem is just a process, and a process is a series of small actions.

The magic chain is the simplest and most effective framework to visualize the process, its actions and the requirements.

For every action in life, you put something into it and get something out of it. The magic chain is made up of actions - the magic. More often than not, we do not need to know how the action actually happens so long as it works.

An action has an input and an output. The input is the resource needed to begin the action. E.g. a bit of information like text or a number, or a physical item like a pen and paper, literally anything! Having no input is in and of itself considered a type of input.

The action takes the input, applies some magic to it and produces an output. Just like the input, the output can be of any form or type, and can be completely different from the input.

Input → Magic → Output

Then we form the connections of the chain. An output of one action is the input to another, and multiple connections create a process - the chain.

Input → Magic → Output → Magic → Output → Magic → Output

Let's go back to Joe's steps and look at them again. You'll be able to see a clear chain appear, so we only need to form it. What does the action need? And what does it produce?

Email, pen, paper → Writing → Draft
Draft → Phone → Digital image
Digital image → Upload → Digital image in Google Drive
Digital image in Google Drive → Download → Digital image on the laptop
Digital image on the laptop → AI → Text
Text → Create file via Obsidian → Text file
Text file → Editing in Obsidian → Boss is happy

The chain makes the entire process and its actions from start to end crystal clear.

The fourth step is to name the pains.

You do that by looking at what you already produced: the story, the steps, the chain, and asking yourself: Where is the pain? What takes too much time? What is annoying? What do you hate doing in that process? What is repeated?
Look at the chain, notice there are 3 steps just to move the image from the phone to the laptop.
The editing sucks because of the editor.
There are 2 actions to transcribe each image.

So, what is the solution?

"The formulation of a problem is often more essential than its solution, which may be merely a matter of mathematical or experimental skill. To raise new questions, new possibilities, to regard old problems from a new angle, requires creative imagination and marks real advance in science." - Albert Einstein

If you did all the steps, you probably already know the solution and what to fix. Finding and defining the problem well - is the hardest and most important part.

Write down the ideal process you wish you had. Do not limit yourself. Imagine the software you think will solve the problem. You don't have to imagine each button and how it looks, just what you want it to do, based on the pains.

Joe's problem is the process because it is unnecessarily long. He adjusted himself to the apps and what the market offers. His process, from a request to an edited piece, could be optimized significantly.

What if Joe could create his own personal software and automate the whole process? Making it take 2 minutes instead of 30, and, while he's at it, replacing his "not-so-liked" text editor with something that is tailored for his process?

All he wants is a simpler editor than Obsidian - a file explorer and basic text editing capabilities. He wants to ease the whole paper-to-digital process, and with a click of a button - transform the image to text.

What is left to do is to clean the wish and form a request. E.g.
I want to create an app, with the following requirements:
A simple text editor with bolding, headers, list capabilities, and a file explorer.
A way to easily share images from a phone to a laptop.
With a click of a button, transcribe the image to text.

This is no longer fiction for Joe. He can leverage AI and ask it to build the app for him. He can do that today and so can you with any challenge you face.

Notice the request is the prompt. Feed it into AI and let it cook. You can ignore the jargon, and just ask AI to build your personal software. Ask it to run the personal software, and be amazed by what comes out.

Once you have the initial app, continue, and do not stop here. Refine and improve, iterate and make sure the process is ideal. Maybe you do not like how it looks? Ask it to improve the design - literally. Do not overthink it! Tell it what you do not like, what to change, what to move, colors and sizes. In your own words. Build your personal software.

Be like Joe, solve your own problems

Joe is me. I needed a solution to the long, annoying process, so I didn't wait for someone else to hand it to me on a silver platter. I knew no one would build it: the problem was mine alone.

I started with the tools I knew and adjusted my process to them. I hated it, so I solved it. I built Writor, a simple text editor that makes it easy to go from pen and paper to digital. As a bonus, using my own software surfaced new challenges, so I just asked AI to add the solution.

As a software engineer with 10+ years of experience, I tell you one thing right now and you better believe me: You do not need to know how to code in order to create your personal software. Ignore what all of the "professionals" tell you!

There is a tectonic shift in code. Leverage it to your advantage!

I did not pay attention to the code that runs Writor at all - I do not need to. It does the job and it works well enough for me. That is the only thing that matters!

All you need to do to start is to spot and define the challenge - the pain. Pay attention to it, look for it - you are constantly surrounded by pains like it and now you can solve them, in ways you never imagined before.

If you want help to start, feel free to contact me. I'd be happy to.

I just wanted to convert an image to WebP and cut a few seconds from a video

Sagi h — Thu, 30 Apr 2026 21:06:00 +0000

I just wanted to convert an PNG to WebP and cut a few seconds from a video! so I built FastEdit

This started with a very stupid problem.

I had a simple media task:

convert PNG to WebP
trim a short video
compress a file enough so I have good light house score
crop something before uploading

The kind of thing that should take 30 seconds.
Instead, every option was bad in a different way.
One site wanted me to upload the file.
Another had ads everywhere.
Another let me edit the file, then asked me to pay when I tried to export.
Another added a watermark.
Another technically worked, but the UX felt like it had not changed since 2009.

And then there are the "proper" tools.
Install a desktop app. Open a full editor. Figure out export settings. Or use ffmpeg and start searching for the right flags again.

All of that is fine when you are doing serious editing.
But most of the time, I was not doing serious editing.
I just wanted to fix the file and move on.

So I built FastEdit:
https://fastedit.net

FastEdit is a free browser-based image, GIF, and video editor/converter.

The goal is simple:

Local
no account
no watermarks
no random export limits
no "subscribe to download"
no bloated editor for tiny file chores

It runs locally in the browser using WebAssembly, so the file stays on your device.

What it does

FastEdit is for the small media tasks that show up right before you post, ship, send, or publish something.

Things like:

Converting images
MP4 to GIF
GIF compression
image compression to a target size
crop and resize
trim video
add text or logo overlays
blur or redact parts of a clip
create favicons
resize thumbnails and social images

Basically, the annoying "last-mile" file chores.
I do not want to open Photoshop for this.
I do not want to install another app for this.
I do not want to upload private screenshots, client assets, or product recordings to a random converter site just to change a format or cut three seconds from a clip.

I want:

Open the page
Drop the file
Fix the thing
Export
Leave

That is the whole product philosophy.

Why local-first matters

A lot of online converters and editors are somewhat convenient for one task, but they usually require uploading the file.

For many files, that is probably fine.
But sometimes the file is:

a client screenshot
a private screen recording
a product demo before launch
a bug report with user data
an internal image
something you simply do not want copied to another server

FastEdit loads the app, then processes the file in your browser.
That means the actual media work happens on your machine.

No upload queue.
No remote worker.
No server copy to delete later.
Just your browser doing the work.

What I am not trying to build

I am not trying to replace professional creative tools.
If you need advanced video editing, use a real video editor.
If you need serious image manipulation, use a real image editor.

FastEdit is intentionally for the boring stuff.
The little tasks that interrupt actual work.
The file is too big.
The format is wrong.
The GIF needs a caption.
The clip has dead air.
The screenshot has something private in it.
The platform wants a weird size.
The avatar needs to be under 256KB.
Those tasks should not require a full editing workflow.

Current features

Right now FastEdit supports:

image, GIF, and video conversion
basic video effects
compression
crop and resize
video trimming
text overlays
image and logo overlays
blur and redaction
batch processing
target file size exports
platform presets for things like Discord, Telegram, YouTube, Instagram, GitHub, favicons, and more
local browser-based processing

It is completely free right now.
No account.
No watermark.
No paid export button hiding at the end.

Tech side

The core idea is to push as much processing as possible into the browser.

The app uses browser APIs and WebAssembly-based processing so common conversion and editing workflows can run locally.

That gives a few benefits:

fewer privacy concerns
no upload wait
no server-side processing queue
no per-file cloud processing cost
works well for quick edits
easier to offer for free

The tradeoff is that performance depends on the user's device.
Big files are limited by browser memory and local CPU power.
But for the kind of small, annoying file tasks I built this for, that tradeoff feels worth it.

Feedback wanted

I would love feedback from other developers and indie hackers.

Especially on:

Is the positioning clear?
Would you use this for README GIFs, product screenshots, demo clips, favicons, or launch assets?
What small media task still annoys you?
Does the no-upload angle matter to you?
Is anything confusing in the first 30 seconds?

You can try it here:

https://fastedit.net

I built it because I was tired of needing five different sketchy websites or one giant overpowered app for basic media chores.

Trying to make the boring file-fixing part of shipping things suck a little less.