In the AI world, everything is supposed to be fast.

For websites, that is mostly true. Idea to live site in a weekend. AI writes the code, I deploy, it is online.

Apps are not that.

I published my English learning app, T Wala English, to the Google Play Store as a solo developer using AI coding tools. The building part was genuinely fast, faster than I expected. Everything after building is where the months went.

This is the honest version: what it cost, how long each stage actually took, and every wall I hit, including the ones no tutorial warned me about.

What It Actually Cost to Start

One line item. One time.

A Google Play developer account costs $25, paid once, and it is yours permanently. No annual renewal. No revenue share on that fee.

Compare that to Apple charging $99 every year and it is obvious why most independent developers start on Android.

That $25 is the entire financial barrier to entry. Everything expensive after that is measured in time, not money.

The Stack, and What Each Part Cost

Most posts about building an app stay vague here. I will be specific.

GitHub Copilot is the AI layer, and the only paid subscription in the whole stack. It is not a chatbot I copy and paste from, but autocomplete living inside the file I am editing. For someone who is not a full time mobile developer, this is the difference between shipping and abandoning the project.

VS Code is the editor and it is free. Nothing exotic is needed.

Expo CLI is what makes React Native approachable, and the free tier is enough. Save a file, your phone updates instantly. That feedback loop is what makes solo development feel possible. It is brilliant, right up until you add ads.

Expo EAS Build produces the actual APK in the cloud. The free tier works, but builds sit in a queue, so plan your day around it rather than waiting at the screen.

GitHub is free and turned out to be the glue. It holds the code and automatically triggers the build, which is what eventually made the whole pipeline hands off.

Google Play Console is the $25 account, and it is where the real work lives. Not the code, the paperwork.

AdMob and AdSense are both free, and you need both. That surprised me, and I explain why below.

If you are building a lean stack for other parts of your business, the same principle applies to no code tools for freelancers and AI automation workflows. Fewer tools, wired together properly.

With that stack, the app came together fast. This is where AI genuinely delivers on the promise.

Then I tried to publish it.

The Wall Nobody Warns You About: 12 Testers, 14 Days

This is the rule that stops most new solo developers cold.

If you created a personal Google Play developer account after November 13, 2023, you cannot publish to production until you have run a closed test with at least 12 testers opted in for 14 continuous days.

Four things to understand about this rule.

It used to be 20 testers. Google reduced it to 12 in December 2024. Most tutorials and videos still say 20. If a guide tells you 20, it is outdated.

Opted in means installed. Someone accepting your invite but never installing does not count toward your 12.

Organization accounts are exempt. Register as a legal business entity instead of an individual and this requirement does not apply at all. Worth knowing before you create your account.

The 14 days are continuous. The clock starts once your release is approved and 12 testers are in.

Do not look for shortcuts here. Using fake accounts or emulators to hit the number risks permanent suspension of your developer account. Two weeks of waiting is far cheaper than losing everything.

I was a newbie. I created a personal account. So I found 12 real people, got them installed, and waited two weeks.

First Release and the Legal Requirements

Once the closed test cleared, the first production release went through smoothly. But two pieces of paperwork must be ready before you hit publish.

A live privacy policy. This is non negotiable. It needs to be a working URL Google can visit, and it must honestly describe what data your app collects. If you use ads or analytics, that counts as data collection. Do not copy someone else’s policy and swap the name.

Play Console declarations. Data safety forms, content rating questionnaire, target audience, ads disclosure. None of these are hard. All of them take longer than you expect. Answering them carelessly is one of the most common rejection reasons.

First release published. Great feeling.

Then I got curious about monetisation.

The Ads Rabbit Hole

Problem one: AdSense before AdMob.

AdMob is the ad network for apps. AdSense is for websites. I assumed I only needed AdMob. You need both, because your AdMob account ties to an AdSense account for payments. So you go through approval twice, not once. Nobody told me this. I found out by hitting the wall.

Problem two: Expo cannot preview ads.

Remember that instant preview loop that made Expo worth using? It does not work with ads.

Ad SDKs are native modules. They do not run in the Expo development client. The only way to check whether an ad actually renders in the right position, at the right size, without breaking your layout, is to build the full app and install it on a device.

My development cycle went from seconds to this. Change one line of ad placement code. Trigger a full build. Wait. Download the APK. Install on a real phone. Look at the ad. It is wrong. Go back to the start.

Every single iteration. This is the most frustrating part of the entire process, and it is why ads took me longer than building the whole app.

What Actually Fixed It: Hiring One Junior Developer

Most solo founder posts leave this part out, so I will be direct.

I hired one junior developer to work and debug alongside me.

Not to take over. Not a team. One person working in parallel, so that while one build was running, the other of us was already on the next problem.

With build cycles that long, the bottleneck is not skill. It is waiting. Two people waiting in parallel is dramatically faster than one person waiting alone.

It also cut the fatigue, which matters more than people admit. Staring at build logs alone for hours is how solo projects quietly die.

If you are stuck in a slow iteration loop, adding one person is often a better investment than adding another tool. It is the same logic behind reducing manual work using AI. Find the real bottleneck, then attack that specific thing.

Update Two and the Traps That Repeat

I assumed updates would be trivial. Mostly they are, but two things come back every single time.

You re-declare. Adding ads changed my app’s data collection, which means the data safety declaration has to be updated. This is not a one time form.

Your privacy policy updates too. Ads collect data, so your policy must say so. If your declaration and your privacy policy contradict each other, that is a rejection.

Once I knew to expect both, the second update was fast. Build, check, ship.

The Architecture That Made Everything Simple

By this point I stopped fighting the process and started building around it.

I write in VS Code, push to GitHub, which automatically triggers an Expo EAS build. I download the APK, test it on a real device with ads live, then publish through the Play Console.

Everything is wired together. I make a change, push it, and the build happens without me babysitting it.

The lesson is that once every piece is connected properly, the whole thing becomes simple. Even with ads, even though you still rebuild every time. The complexity never disappeared. It just stopped being manual.

For future updates I plan to disable ads in development builds and enable them only in the final production build. That should remove most of the rebuild pain during normal feature work.

Where the App Stands Today

T Wala English is live with ads running, a working UI, a tag system, a revision feature and expanded content.

It is not finished. What is still ahead: adding a real AI feature that is an actual differentiator rather than a thin wrapper, a real backend which means rewriting the privacy policy again, and a senior developer audit before scaling further.

I am saying that plainly because the shipped an app in a weekend genre is mostly fiction. Shipping is not the finish line. It is the point where you find out what you built wrong.

So Is AI Coding Good for Building Apps?

Yes, genuinely. But the promise is oversold, and it is worth being precise about where it helps.

AI made the building fast. Code that would have taken me weeks took days. For a solo developer without a mobile background, this made the project possible at all.

AI made none of the rest faster. The 14 day tester wait. AdSense approval. AdMob setup. Data safety declarations. Privacy policy compliance. Build times. Ad rendering bugs. AI touched none of it, and that is the majority of the calendar time.

Websites are fast because the entire pipeline is fast. Apps have a gatekeeper, a review process, a compliance layer and a native build step.

AI accelerates one stage out of five. Plan for that and you will ship. Assume AI collapses the whole timeline and you will quit somewhere around the tester requirement.

Every Stage, Summarised

Setting up the Play Console account cost $25 once and took about a day. AI was no help, and none was needed.

Building the app cost a Copilot subscription and took days rather than weeks. This is where AI helped massively.

Closed testing was free but took 12 testers across 14 continuous days. AI could not touch it.

The privacy policy was free and took about half a day, with AI helping on the drafting.

Play Console declarations were free and took a few hours per release. No AI help.

AdSense and AdMob approval was free but meant days to weeks of waiting. No AI help.

Ad implementation was free and by far the longest stage, where AI helped only partially because the bottleneck was build time, not code.

The second update was free and fast, once I understood which declarations came back around.

If you are starting today, do these three things. Register as an organization account if you legitimately can, because it skips the 12 tester requirement entirely. Get your privacy policy live early, before you need it. And line up your 12 testers before your app is finished, so the clock runs in parallel with development instead of after it.

Those three decisions would have saved me the most time.

Frequently Asked Questions

How much does it cost to publish an app on the Play Store?
A Google Play developer account costs $25 as a one time fee. There is no annual renewal. Publishing itself is free. Your remaining costs are tools and your own time.

How many testers do I need for Google Play?
Twelve testers, opted in continuously for 14 days, for personal developer accounts created after November 13, 2023. This was reduced from 20 in December 2024. Organization accounts are exempt.

Do I need AdSense to use AdMob?
Yes. AdMob is linked to an AdSense account for payment processing, so you effectively go through two approvals rather than one.

Can I test AdMob ads in Expo Go?
No. Ad SDKs are native modules and do not run in the Expo development client. You must create a full build and install it on a real device to see ads render.

Can I publish an app without coding experience?
With AI assistance, building is achievable. But publishing requires understanding privacy policies, data safety declarations and testing requirements. The code is the easy part now. Compliance is not.

How long does it take to publish an app solo?
Building can take days with AI assistance. Realistically, budget four to eight weeks end to end, including 14 days minimum for closed testing plus review cycles, ad network approvals and iteration time.

Final Thoughts

The AI era genuinely changed what a solo developer can build. It has not yet changed what a solo developer has to survive in order to ship.

If you are building something of your own right now, I write about the systems behind this kind of work, including using AI for business growth and the workflows that keep solo projects moving.

Building an app solo at the moment? Tell me which stage you are stuck at in the comments and I will answer from experience.