I found a social media agency with a static, simple website that I thought could be much better. Instead of writing a teardown or sending a vague pitch, I challenged myself to recreate the site with Opus 4.7 and see how quickly I could reach a result worth offering.

It took four prompting attempts. I then emailed the agency with a direct proposal: $500 to acquire the redesigned site, plus $40 per month to host it with me. They accepted, and when I posted the experiment, the transfer was underway.

I had deliberately kept the test small: a simple target, four iterations, a fixed acquisition price, and a small recurring hosting offer.

There was no discovery phase or open-ended proposal in the version I shared; I put a finished alternative and two numbers in front of the agency.

A narrow site made the experiment measurable

I chose a site that was static and simple. I was not replacing a complex application, migrating a database, preserving a large publishing system, or integrating business-critical workflows. I was recreating a straightforward agency site.

That made four prompting attempts a meaningful unit. I could ask whether the model-assisted process produced a redesign I was willing to put in front of the owner without turning the experiment into an open-ended build. On the fourth try, I thought it had.

“Four tries” refers to the prompting iterations used to recreate the site, not four total actions from discovery to transfer. I did not share the prompts, the editing history around them, the publishing stack, or the time each attempt required. I never measured enough to turn that number into a speed benchmark.

For this static, simple site, I reached a version I considered strong enough to sell after four Opus 4.7 prompting attempts.

I sold a defined offer, not an abstract redesign

The email pitch had a price and a shape. The agency could acquire the redesign for $500 and keep hosting it with me for $40 per month.

Those two numbers separated the artifact from the ongoing service. The $500 covered acquiring the redesigned site according to my offer. The monthly amount covered hosting. I did not include support, revisions, maintenance, domain management, contract length, or any other service in the short description, so I will not attach extra promises to it now.

The email had one job. I had already made the thing I wanted to sell, so the recipient could react to a specific result and a specific price. I did not turn the first message into a broad consulting proposal.

I summarized the outreach in one line. I did not share the complete email, subject line, follow-up sequence, or private conversation, so there is no polished cold-email template behind this story. There was simply a redesign, a purchase price, and a hosting price.

The deal was still in motion

When I shared the experiment, the agency had accepted and the site was being transferred. I did not share final transfer records, payment documentation, or a later performance report. I had selected the site because I thought its static, simple design needed improvement.

The result stopped there: an accepted $500 acquisition offer and a proposed $40 monthly hosting relationship. I did not measure conversion lifts, traffic gains, customer response, long-term retention, or the agency's revenue after the redesign.

I also did not share private client terms or permission details, so I am leaving them out rather than filling in a cleaner sales story after the fact.

Build first, then name the price

The sequence kept me honest. I identified a narrow target, produced a concrete alternative, named a price, and received a yes. Each step forced the next decision.

This one accepted offer is not a market study. It was a specific transaction-in-progress: a simple site, an AI-assisted four-iteration build, a $500 acquisition price, and a $40 monthly hosting offer.

The experiment attracted 316 bookmarks, slightly more than its 278 likes. I cannot know why any individual saved it. What matters to me is that AI reduced enough friction for me to build the speculative redesign before asking for the sale, and the agency accepted the offer I put in front of them.