Code Was Always Cheap

People keep saying code is cheap now because of AI. Anyone can build a small app in a weekend. The gap between idea and working prototype is getting smaller every month.

I understand why people say it. But every time I hear it, I think about 2020.

The product we were building

During covid, I was working at a company that was building custom ERP solutions for retail companies in the Middle East. The product was not some random side project. The team was good. The goals were clear. Everyone was serious about building.

We were shipping one feature after another. Inventory. Billing. Operations. Reporting. Whatever the product needed next, we built it.

There was also a lot of focus on making it scalable. Not in the fake pitch deck way where everything is called scalable because it sounds good. We were actually thinking about real stores, real usage, real data, and real operational problems.

The product had early adopters too. A few supermarket chains agreed to test it for some of their operations. They were not random leads from the internet. Most of them came through the founders’ network in the Middle East.

At the time, that felt like a good start. You have a solid team, a product, and real businesses trying it. You keep building. It feels like progress.

But progress inside the team is not the same as progress in the market.

The part we ignored

That was the part we learned slowly. The whole team was focused on building. There was no real marketing team. There was no serious distribution engine. There was no clear plan for how this thing would reach companies outside the founders’ personal network.

So the product stayed inside that circle. It worked for early conversations. It worked when the founders could introduce it directly. It worked when trust already existed before the sales call. Outside that network, things did not move.

The product failed to gain traction. Not because the team could not build. Not because the code was impossible. Not because the idea was useless.

It failed because building the thing was only one part of the job. The other part was getting it out there. And that part never really happened.

The B2C attempt

We also tried to turn part of it into a multi-vendor application. Retail outlets could sign in and sell their products online. This was during covid restrictions, when everyone was trying to sell online. On paper, the timing looked good.

People were stuck at home. Shops needed online sales. Customers were already changing habits. If there was ever a time for that kind of product to work, it felt like that was the time.

But that failed too. It did not capture the market. Again, not because nobody could write the code. We could build the pages. We could build the flows. We could connect the stores. We could add features.

But features do not magically create demand. Posting on company social media pages is not marketing. Having a product is not distribution. Being useful is not the same as being trusted.

The founders were not very keen on spending money on marketing. They were not very keen on building a serious marketing function either.

The idea of marketing was mostly to post something on the company’s social pages and hope people noticed.

People did not notice. Or if they noticed, they did not care enough. Or if they cared, they did not trust it enough. Or if they trusted it, they still had no strong reason to switch from whatever they were already doing.

The market does not wait

That is the part people skip when they talk about products. They talk as if the market is sitting around waiting for your app.

It is not.

Most people are busy. Most companies are messy. Most teams already have some broken process that works just enough to avoid change. Replacing that takes more than clean code.

It takes trust. It takes timing. It takes sales. It takes positioning. It takes someone repeatedly explaining why this product matters until the market finally understands it.

That is hard. It was hard before AI. It is harder now.

What AI changed

When people say code is cheap in the AI era, they are not completely wrong. AI has made building faster. It has made prototypes easier. It has reduced the cost of trying an idea.

But code was already cheap in a lot of markets. Your niche B2B SaaS idea could already be built by another company. Your custom dashboard could already be copied. Your workflow tool could already be recreated by a team with enough developers and time.

The hard part was never only the code. The hard part was making the right people know it exists. The hard part was making them care. The hard part was making them believe it will still be around next year. The hard part was convincing them to change how they work.

AI did not create that problem. It just made it more obvious. Now more people can build, more people can ship, and more people can put a decent looking product on the internet.

So the market gets louder. And when everything gets louder, being heard becomes more expensive.

The actual lesson

That is what 2020 taught me. Not that building is useless. Building matters. Good engineering matters. Reliable software matters.

But building alone is not a business.

Code was cheap before AI too. The expensive part was always attention, trust, and distribution.

Now it is just easier to forget that because the demo looks better.