Phoenix's Startup Journal

I Can't Code. I Shipped 36 Versions in Six Days.

2026-10-31T00:00:00Z

Six days ago I was reading PDFs on my laptop and getting more and more annoyed with the app I was using. A little before midnight I sent Claude a message. It started like this:

I want to build a PDF reading tool for myself… I'm not a programmer, so please talk to me in simple language.

Today the tool is on version 36. It reads PDFs, e-books and Markdown files. It translates as you read, lets you ask AI about any sentence, and turns your notes into key points and a mind map with one click. There's an international edition and a separate edition for mainland China, each with its own website. A few friends are using it to read papers and books, and when something bugs them they can complain about it right inside the app.

In those six days I didn't write a single line of code.

I want to write this down. The tool isn't anything remarkable. But building it showed me that the barrier to making a product has moved.

The hard part is no longer "how". It's "what".

Over those six days I sent Claude more than 170 messages. Looking back, the thing I said most often was some version of "don't build it yet, confirm the requirements with me first." I said it more than thirty times.

My very first message ended that way: "Please don't write any code yet. First give me a plan: what technology to use, how the data will be saved, what the steps are. Then ask me whatever you need to confirm."

I didn't plan to work like this. It became the rhythm almost immediately. I'd describe what I wanted, the AI would turn it into a concrete plan, and then it would ask me questions. Where does this button go? What happens in that case? Should there be a setting to turn it off? Only after I answered would it start building. Many of those questions had never occurred to me. Being asked them showed me how vague my "I want" really was.

It used to be that when someone who couldn't code had an idea, what stopped them was how. Now "how" is barely a problem. What's left is all what: what problem you're actually solving, what good looks like, and what not to build. AI can't answer those for you. But it's very good at making you ask them properly.

I was the first user

The tool improved fast for a simple reason: I used it every day.

Whenever something felt awkward, it became a new version that same day. At first you had to select text and then click "Highlight". Two steps felt like one too many, so now dragging across a sentence highlights it. When I read English with translation on, the translated page would often jump to the next page before I'd finished reading, and I'd lose my place flipping back and forth. So the translation now flows continuously, one page after another.

None of these are big features. But they decide whether something is merely usable or actually pleasant to use. And you only notice details like these if you really use the thing every day.

What I figured out reading Paul Graham

On the second night I was reading Paul Graham's essays, and somewhere in there I realized I wasn't really building a PDF reader.

A lot of people read books, especially business and self-improvement books, and feel inspired while reading. Then they find it very hard to actually use any of it in their lives. The knowledge gets into their eyes but not into their heads. What I care about was never remembering what I read. It's digesting it until it becomes mine.

So what the product really needs is a whole loop. You note your thoughts while reading. When you finish, AI organizes them into key points and a mind map. Then it quizzes you on your own notes. Later you come back and check whether you really absorbed it. That night I gave myself a rule:

Every feature should be designed around one goal: helping people digest what they read and put it to use.

Since then, whenever I want to add something, I ask myself first: does this help someone digest what they read, or is it just one more button?

What my first real user taught me

On day five, a friend who's doing a PhD used the tool on an ordinary morning to read a paper from start to finish. He wasn't testing it as a favor to me. He actually had something to read that day. That made me really happy.

He gave me a lot of feedback, and one point changed how I understood the product. He wanted literature summaries. If he could train this agent to summarize papers the way he wants, he said, it could replace ChatGPT for him.

Talking it through, we realized that most people don't read PDFs for fun. They read them because they have to: papers, reports, work documents. For them the most honest need fits in one sentence: "I don't want to read this, just tell me what it says."

So that sentence became the name of a button. And we can do it better than pasting the text into ChatGPT, because every point in the summary comes with the original sentence and page number. One click takes you back to that sentence in the document, where you can highlight it or save it to your notes, with no copying and pasting between two apps.

He also told me what he didn't like. For example, he thought the quiz feature was pointless. I kept it. People who don't need it can simply not use it. But his feedback made one thing clear: some people open the same tool to read closely, and others just want to know quickly what's inside.

The real barrier wasn't code

People assume the hardest part of building a product without knowing how to code is the code. Over these six days, what gave me the most headaches was everything around it.

I wanted friends to try the tool, but I couldn't hand them my AI key. That meant invite codes and a relay server. I had never used Cloudflare, so I had the AI walk me through it click by click. Claude isn't available in mainland China, so I made a separate mainland edition that runs on MiniMax, a Chinese model, with its relay on Alibaba Cloud. There I stared at error messages and settings I had never seen before and sent the AI screenshot after screenshot. Then, so friends could just open a link, I bought a domain and learned what DNS records are. Late at night I sat watching a Tencent Cloud dashboard, waiting for "Deploying" to turn into "Active".

None of that was code. It was accounts, permissions, networks and regional restrictions, the friction of the real world. AI could explain every step. But I still had to click every button myself.

The hardest part is saying no

Once AI can build whatever you ask for, what's scarce changes.

Several times a plan for a new feature came with a setting that let users switch it on or off. I took those settings out. Translation follows you automatically when you turn the page, with no setting. Opening the app takes you straight back to where you left off, with no setting. I want it to feel as smooth as possible, not cluttered with settings no one needs.

By day six the tool did a lot, and the toolbar had become bloated. I've always believed in Steve Jobs' idea that simplicity is the key. So we made a minimal mode. When you open a book you see only the essentials, and everything else lives under "More". People reading research papers can switch to a pro mode with one click. Then we redesigned the whole interface from scratch.

Adding a feature now costs almost nothing. So the real work becomes deciding what not to add.

Why this site exists

These six days convinced me that one person with an idea and AI can now get surprisingly far. I'm going to keep building this product, and I want to write down how it goes: what I did each day, what feedback I heard, what I decided, and what I figured out.

You'll find two kinds of writing here. Essays, like this one, are about things I've figured out. The log has one entry per day and covers the day-to-day of building a startup. The site is an experiment too. AI will draft most log entries from my day's work, and I'll edit them and add my own thoughts. It's the same way I built the reader: I work out what I want, and AI builds it.

If you're building things with AI too, or you're curious about the reader, I'd be glad to see you here again.

Day 6 · Simplicity is the key

2026-10-31T00:00:00Z

Today I went from v0.27 to v0.36.

From morning to afternoon it was one small improvement after another, mostly from things that bugged me while reading. The cursor turns into a pen over the text. Notes fold themselves into small sticky notes once you're done writing, so they don't block the page. Translation now works in 19 languages. You can copy selected text with one click. A single click in the table of contents bookmarks a chapter. And one thing had been bugging me for days: with translation on, the translated text would jump to the next page before I'd finished reading. Now the translation flows continuously, one page after another.

In the evening I made a bigger decision. The tool did plenty by now, but the toolbar had become bloated. I've always believed simplicity is the key, so I redesigned it in three steps:

A minimal mode and a pro mode. Books open in minimal mode with only the essential buttons. People reading research papers can switch to pro mode with one click.
One clean, Apple-like look across the whole app, with dark mode.
A new home screen, bookshelf and notebook.

At night I started building this website. It went live the same night, and I set it up so that AI drafts a log entry every day.

What I figured out today: adding a feature now costs almost nothing. The real work is deciding what not to add.

Versions today: v0.27 pen cursor, sticky notes · v0.28 no pen when auto-highlight is off · v0.29 translate into other languages · v0.30 Copy button, one-click chapter bookmarks · v0.31 continuous translation · v0.32 bookshelf refreshes itself · v0.33 highlight the original inside the translation panel · v0.34 minimal / pro mode · v0.35 unified visual design · v0.36 new home screen, bookshelf, notebook

Day 5 · "I don't want to read this, just tell me what it says"

2026-10-30T00:00:00Z

Today I got the first feedback from someone using the tool for real.

A friend who's doing a PhD read a paper with it this morning, start to finish. He wasn't testing it for me. He actually had to read the paper. He said the translation works really well and he figured everything out without reading any instructions. He also found problems. When he wanted to select a sentence to ask AI about it, it got highlighted instead. The proportions of the Ask AI panel felt off. And he wanted to ask about the whole paper, like "what's the main contribution here?", without selecting anything.

The most important point was that he wanted AI to summarize papers for him. Talking it over, we realized most people don't read PDFs for fun. They read them because they have to. So now there's a button called "I don't want to read this, just tell me what it says". Every point in the summary links back to the original sentence, and that's where we beat pasting the paper into ChatGPT. A friend's younger sister wanted to read Markdown files, so I added that too.

I started keeping a requirements sheet that records every round of feedback and what we decided. I shipped ten versions today. Translation in the China edition now streams in paragraph by paragraph, which is much faster. The website updates itself whenever I push to GitHub. Opening a personal invite link sets up the AI automatically. Translation follows you when you turn the page. The app reopens right where you left off. There's also "Roast it", where users can send feedback anytime and see which update their feedback turned into, and a "Recommend to a friend" button.

What I held the line on today: no pointless settings. I want it to feel as smooth as possible.

Versions today: v0.17.2 faster China translation · v0.18 automatic website deploys · v0.19 personal links · v0.20 version number, what's-new notes · v0.21 whole-document Ask AI, "just tell me what it says" · v0.22 AI sorting for the bookshelf · v0.23 Markdown files · v0.24 auto-translate on page turn · v0.25 Roast it, Recommend to a friend · v0.26 pick up where you left off

Day 4 · Sharing it: invite codes, a China edition, and a domain

2026-10-29T00:00:00Z

Today was about one thing: how to let other people use it.

International edition: I added invite codes. A friend gets a code and can use the AI features without ever touching my key. Behind it is a relay server on Cloudflare. I'd never used Cloudflare, so Claude walked me through it click by click (v0.14).

Mainland China edition: Claude isn't available in mainland China, so this edition uses MiniMax, a Chinese model, with its relay on Alibaba Cloud (v0.15, v0.16). This was the most painful part of the day. I couldn't find a setting I needed, then hit a 403 error, and sent screenshot after screenshot until it finally worked.

In the afternoon I wrote an intro to send friends and made a step-by-step picture guide for entering the invite code. I was worried people wouldn't know where to paste it.

In the evening I thought: sending a zip file that people have to unpack is too much hassle. Why not just send a link? So the tool became a website (v0.17). I bought a domain and set up a subdomain for the test version. The China edition is served through Hong Kong on Tencent Cloud. Late at night I sat waiting for the status to change from "Deploying" to "Active". Then it said Active and still wouldn't open. It finally did, sometime after midnight.

What I learned today: the hardest part of building a product often isn't the code. It's the accounts, permissions, networks and regional restrictions.

Versions today: v0.14 invite codes (international) · v0.15 China edition (MiniMax) · v0.15.1 China translation fix · v0.16 China invite codes · v0.16.1 China relay live · v0.17 web version · v0.17.1 China website

Day 3 · My first user, and eight versions in a day

2026-10-28T00:00:00Z

I shipped eight versions today (v0.6 to v0.13).

Right after midnight I added EPUB e-books. Then came red, yellow and blue highlights, each color with a name you choose (like "look this up"), plus saving AI answers and a reading timer.

Around noon I started looking for my first users to test the idea. The first one quickly sent me 13 pieces of feedback: remember the last color I used, let me bookmark favorite chapters, let me highlight inside AI answers, generate a mind map on export, make an iPad version… Whatever I could do, I did the same day. Anything that needed more information went on a list.

In the afternoon I added an English interface and "Translate". Chinese is my first language, so when I read English papers, one click translates the page paragraph by paragraph, with the original and the translation side by side. Then I started keeping a changelog and put the code on GitHub, so every version is on record from now on.

In the evening came bookmarks, a help button, and a built-in tutorial that replaces the user manual, and replaces me explaining things over and over. The bookshelf now looks like a real one, with covers facing out on wooden shelves. Last came a notebook that collects the notes from every book in one place.

Somewhere in the middle I ran out of usage once and had to wait a while before continuing.

The real problem I hit today: to let anyone else try the tool, I'd have to give them my own AI key. And Claude isn't available in mainland China. Both problems are for tomorrow.

Versions today: v0.6 EPUB · v0.7 three highlight colors, reading timer · v0.8 first user's feedback · v0.9 English interface, Translate · v0.10 bookmarks, help, tutorial · v0.11 visual bookshelf · v0.12 resizable translation divider · v0.13 notebook

Day 2 · It's more than a reader

2026-10-27T00:00:00Z

I only worked for a bit in the evening today, but I did more thinking than on any other day.

First, one small change. Selecting text and then clicking "Highlight" was two steps too many, so now dragging across text highlights it right away (v0.5).

Then I started wondering: what if this became a product people pay for? Something like a subscription "second brain" for heavy readers. AI would organize everything you've read, sort it into topics, shape it into something like a mind map, and quiz you now and then, a bit like Quizlet. Claude and I brainstormed for a long time.

Later, reading Paul Graham's essays, one thing became clear. Many people read books, business books especially, and feel inspired while reading, but find it hard to apply anything afterwards. What this product should really solve is helping people digest what they read and actually use it.

The full loop should be: capture your thoughts → AI summary → mind map → quiz based on the mind map → check later whether you really learned it. I want all of it in one product, as complete an experience as an Apple product.

The rule I set today: every feature should be designed around helping people digest what they read.

Versions today: v0.5 drag to highlight

Day 1 · Version one: give me a plan first

2026-10-26T00:00:00Z

A little before midnight the night before, I sent Claude my first message. I wanted a PDF reading tool just for me that opens in the browser and needs no login. I listed what version one should do: highlighting, attaching a thought to a sentence, dropping a free note anywhere on the page, saving automatically, and exporting everything with one click. At the end I added: don't write any code yet, give me a plan first.

Claude came back with a plan and seven questions. Less than half an hour after I answered them, version one worked.

During the day I moved my notes folder into Google Drive. I also made exports go to a folder I choose, and put together a zip file I could send to other people, with a short how-to guide.

In the afternoon I asked a question: when I export, could AI organize my notes automatically? That was the first time I connected AI to the tool. I got a Claude API key and learned that the models differ a lot in price, and the cheapest one is plenty for tidying up notes.

By evening there was a bookshelf, "Ask AI", a quiz mode and topic summaries. I also set up a reading knowledge base in Notion along the way.

What I figured out today: I don't care about remembering what I read. I care about absorbing it into my own knowledge and slowly building a knowledge system of my own.

Versions today: v0.1 first version · v0.2 choose where exports go, share with others · v0.3 export + AI digest · v0.4 bookshelf, Ask AI, quiz, topic summaries