I Built the Attention Lasso Website for $35.33.The Real Cost Was the Learning Curve.

I registered Attention Lasso as a company on 19 January 2025. Then I gave it the full founder treatment: I admired the certificate, thought about the future and went back to my job.

I was still leading digital growth at Quick Med Cards. Attention Lasso existed, but mostly inside my head and a folder full of experiments. Every so often, I would search for the domain to make sure nobody had bought it.

This was not market research. It was anxiety with a search bar.

What if somebody bought attentionlasso.com first? Imagine registering a company, taking your time, then meeting a stranger who would like to sell your own name back to you at the price of a small Toyota.

In September 2026, I finally bought the domain and launched the business. The website cost me $35.33 in cash to get live:

• $12.13 for the domain through Cloudflare Registrar
• $23.20 for one month of ChatGPT Plus
• $0 for hosting on Netlify

That total includes $4.87 in VAT. It does not include my time, my existing skills or the many conversations I had with machines that were extremely confident and occasionally wrong.

The low price is the headline. The more useful story is what happened between buying the domain and having a website I could confidently attach to my name.

I was launching a company that promised to help businesses get noticed. I could not sell visibility from a website that disappeared into the wallpaper.

The website existed many times before it existed

I did not begin with a blank page. Long before the launch, I had been using Claude, ChatGPT and other language models to turn ideas into HTML and CSS mock-ups.

Some versions leaned into Attention = $$$. Others announced that the future of marketing is cyborg. One looked like a serious technology company. Another looked like marketing had consumed three espressos and discovered neon.

The experiments were not just decoration. Each one helped me test the questions that actually decide whether a website works:

• Can a visitor understand the offer within a few seconds?
• Does the page speak to a real customer problem?
• Is there proof behind the promise?
• Does every section lead naturally towards a decision?
• Does the brand feel distinctive, or does it look like an AI wearing our logo?

By the time I was ready to build, I knew the visual direction, the tone and the job of the homepage. I had also rejected enough attractive nonsense to know that a polished mock-up is not the same thing as a commercial website.

Qwen produced the beautiful wrong answer

Then I tried Qwen. It produced a design that was completely different from the one I had planned.

The irritating part was that it was good. Very good.

For a moment, I was ready to abandon the plan and follow the shiny new idea. This is one of AI's more dangerous talents. It can produce an exciting alternative before you have finished remembering why your original decision existed.

I kept the design for reference, admired it and returned to the brief. The customer did not need to be impressed by how many visual directions I could generate. They needed to understand what Attention Lasso could do for their business.

Good design attracts attention. The right design directs it.

Grok built 29 pages before the free usage ran out

Grok changed the project. I gave Grok Build the design direction I had already tested, and it turned that direction into a working website.

Not a homepage. A 29-page website.

For a free tool, that was absurdly impressive. It was also an excellent way to feel productive before inspecting all 29 pages.

The free usage ran out before the build was fully finished. What remained was a website with strong visual energy, incomplete details and copy that had the nutritional value of cardboard. It sounded professional in the way every business sounds professional when it says it delivers innovative, customer-focused solutions.

The same copy could have belonged to an agency, a fintech company or an ambitious car wash.

That was a serious problem. Attention Lasso was supposed to have a point of view. The website could not merely occupy space on the internet. It had to demonstrate that we understood attention, persuasion and the journey from curiosity to action.

Grok had supplied a remarkably strong first build. It had not supplied the final judgment.

I asked for a website and received a new technology stack

My comfort zone is HTML and CSS. I understand the files, make the change, refresh the page and carry on with my life.

The project I downloaded from Grok had other plans.

I initially thought it was built with Next.js. After inspecting the dependencies, I discovered that it was a React codebase using TanStack Start and Vite. That is a capable modern setup. It was also not something I could casually edit without understanding how the pieces connected.

Opening the project felt like asking for directions and receiving the architectural plans for the entire city.

This exposed a mistake that many business owners will not notice until launch day: the best website stack is not simply the cleverest one. It is the one your business can maintain. If changing a sentence requires finding a developer, opening a terminal and saying a small prayer, the website is going to age badly.

I moved the project into ChatGPT Work and started cleaning it up. We repaired layouts, refined the copy, corrected pages and removed the “Made with Grok” tag. Grok had already done plenty. It did not also need a billboard on my footer.

AI accelerated the production. I still had to understand the business, challenge the output and decide what deserved to remain.

The design was good. The SEO had entered witness protection.

Visually, the website looked ready. Search engines, unfortunately, do not rank a page because the founder likes the colour palette.

Once I reviewed the site as an SEO rather than an excited founder, the problems became obvious:

• Pages shared repeated title tags
• Meta descriptions were weak or duplicated
• Some H1s did not match the purpose of the page
• Several pages existed without a clear search intent
• The copy used broad agency language instead of answering buyer questions
• Internal links did not create a useful journey through the site
• Important pages lacked the structure needed by search engines and AI systems

This is the hidden risk of AI speed. Creating 29 pages quickly feels like progress. Discovering that the pages do not know why they exist feels like being punished for your own efficiency.

Google recommends descriptive page titles and meta descriptions, including unique descriptions for individual pages. Those recommendations are basic for a reason. A page needs a clear subject before Google can confidently decide when to show it.

I mapped the important pages to customer intent. Titles and descriptions were rewritten to earn the click rather than fill a field. The heading structures had to guide both the reader and the crawler. Internal links needed to connect related services and ideas instead of behaving like decorative blue furniture.

I also reviewed how the business, services and author were represented for machines. Schema can clarify entities and relationships, but it cannot rescue vague thinking. If the page does not provide a useful answer, adding more JSON-LD only gives the confusion a name badge.

This is where my experience mattered most. The site did not only need to look finished. It had to be technically understandable, commercially persuasive and discoverable when a potential customer searched on Google or asked an AI tool for a recommendation.

What the website actually cost

The invoice was short. The work was not.

OpenAI lists ChatGPT Plus at $20 per month before applicable taxes. Cloudflare says its registrar sells domains at cost without adding a markup. My receipts reflect the 16% VAT charged to me.

The honest cost was higher than $35.33 because my time was not free. I brought years of SEO, content and conversion experience to the project. I wrote the direction, judged the designs, reviewed the architecture and rewrote the important pages. I also had to learn enough about an unfamiliar codebase to avoid breaking the website while fixing a heading.

Still, the cash cost matters. AI made it possible to create a capable foundation without beginning with a large development budget. It reduced the price of production, but it did not remove the cost of judgment.

That is the part the “build a website in five minutes” videos tend to skip. Five minutes can produce a page. A business website begins with the decisions that follow.

Free hosting met a founder who kept deploying

Netlify hosted the website for $0. Then I began making improvements.

One corrected heading needed a deployment. A spacing adjustment needed another. Then I noticed a mobile issue, an awkward paragraph and one button that had apparently chosen its own career path.

Tiny fixes travel in packs.

Eventually, I exhausted the free credits available to my account. Netlify's credit-based plans include a monthly allowance, and projects can pause when a free account reaches its limit.

The lesson was not that free hosting is a trap. The lesson was that an unplanned review process can make free resources disappear surprisingly quickly. Approve the copy first. Group the corrections. Test locally. Then deploy a version worth reviewing.

I learned that lesson shortly after treating the deploy button like a light switch.

What I would do differently

I would still use AI. I would simply give it stronger boundaries before asking it to move at full speed.

• Write the brief first. Define the audience, offer, proof, tone and primary action before generating a design.
• Plan the sitemap before the pages. Every page needs a customer purpose and a search purpose.
• Choose a maintainable stack. The website should not become untouchable the moment the tool that created it leaves the room.
• Build one approved template before multiplying it. Finding a problem on one page is quality control. Finding it on 29 pages is a lifestyle.
• Write unique metadata from the start. Duplicate titles are easier to prevent than to untangle later.
• Plan for a CMS. A business owner should not need an AI coding session every time a comma needs moving.
• Batch reviews and deployments. Free credits deserve a calmer life than mine had.
• Treat AI output as a draft. Speed gets you to the first version. Expertise gets you to the useful one.

You do not need to know everything before launching. I certainly did not. You do need enough control to improve the website once customers, search engines and reality begin testing it.

Your business does not need my exact website

It needs the same clarity.

Perhaps you have delayed getting online because an agency quote required a board meeting. Maybe you tried a website builder and spent three hours deciding whether the button should be slightly rounder. Or the site is already live, but it receives the digital equivalent of one confused goat walking past.

The problem is rarely “we need some pages.” A useful business website must explain the offer, earn trust, guide action and remain easy to update. It also needs to give Google and AI tools enough clear information to understand when the business is relevant.

Attention Lasso can help you build that system without turning the process into a six-month archaeological dig. You remain the expert in your business. We make that expertise easier to see, understand and choose.

This is only Part 1

Getting the website live was not the finish line. It was the moment the measuring started.

The next articles in this series will cover:

• How I set up Google Analytics 4
• How I connected Google Search Console
• The frustration of launching to zero visitors
• What happened when Google did not index the pages quickly
• How I set up a professional business email
• Why I need a CMS so every edit does not become a development task
• How the website changes once real search and visitor data arrive

I will show the wins and the awkward parts that polished case studies normally hide. The website began with $35.33, twelve experiments and a mild fear that somebody would steal my domain.

Now the real work begins.

Keep pulling the thread

Related reading

Build With Attention Lasso

How I Got a Professional Business Email for Free

Read next

Strategy

What Is Attention Engineering?

Read next

Search Architecture

SEO, AEO and GEO: The Difference That Matters

Read next