I used Codex to build six landing pages for Local Service Spotlight and publish them on Whop. The pages bring our Power Hour, agency retainer, Knowledge Panel package, and AI Builder program into one place, with a separate page for partners. We kept the work in our own source files and used Whop’s command-line tools to upload it.
The result: six pages are live on Whop, with direct checkout links for Power Hour and AI Builder, qualification paths for the services that need an assessment, and a working offer selector.
Why did I pay $50 for Whop AI when I already had Codex?
I had asked my assistant to help move more of our offers onto Whop. During that work, it directed me to buy $50 of Whop AI credits. I paid, then asked why we needed another AI builder when I already had AI accounts available.
That question changed the approach. Codex could write the site and use Whop’s CLI to upload it. We should have exhausted that route before asking me to buy more builder credits. The assistant acknowledged the mistake and finished the build without further Whop AI use.
Whop AI remains an option. For this job, our existing coding assistant was sufficient. Hosting, checkout, payment processing, and AI generation are separate capabilities; choosing Whop for commerce did not require us to use its AI for every page edit.
This was not a zero-cost experiment. I had already bought the $50 credit package, we used an existing AI account and a GitHub build runner, and Whop’s applicable hosting and commerce charges still matter. The useful lesson is to check what you already have before buying another tool.
What did we actually build?
I wanted buyers to understand which kind of help they need. A focused working session, an ongoing agency engagement, personal-brand work, and an apprenticeship are different purchases. Each deserves its own explanation and next step.
| Page | What it does | Buyer’s next step |
|---|---|---|
| Offer hub | Compares the four offers and includes an interactive recommendation selector. | Choose an outcome and open the relevant offer. |
| Power Hour | A 60-minute working session with me for $1,500, including the recording, prompts, completed configurations, and a one-page prescription. | Pay through Whop, then complete preparation and scheduling. |
| Home Service Growth System | Our $3,500/month agency retainer connects Maps visibility, proof and content, and calls and conversion. | Start with an audit and agreed written scope. Ad spend is separate. |
| Knowledge Panel package | Separates $2,500 Claim/correct work from a $7,500 Standard Build. Contested names need a quote. | Get an assessment before enrollment. |
| AI Builder program | A one-year apprenticeship with real client work, priced at $7,500 in three monthly payments of $2,500. | Review the program and its existing Whop checkout. |
| Partner page | Explains referring a buyer to our offers and referring a business to Whop. | Request the appropriate affiliate link or discuss a network rollout. |
Prices describe the offers at publication on September 6, 2026. The linked offer pages explain the current scope and terms.
The details matter. AI Builder has three installments, not an endless monthly obligation. A Knowledge Graph ID or thin card is an early entity milestone, not the same result as a full Knowledge Panel. Google controls panel appearance. And paying for a Power Hour does not automatically reserve a calendar slot.
How did we build the pages without more Whop AI?
Codex wrote the page content, styling, navigation, and small interactive features into a source project. We used the existing offers as the commercial references and put a real photo on every page. We also linked to the underlying methods so someone could understand the work before buying.
The Mac blocked native tools needed to package the site. We moved that compile step to a separate branch in our existing GitHub repository. The build runner compiled and checked the source; it did not receive our Whop credentials or call an AI model.
After the checks passed, the authenticated Whop CLI uploaded the finished package and its source. We reviewed the hosted draft, then promoted the reviewed version after I approved publication. The official Whop CLI repository is the technical starting point for someone who wants to use the same approach.
What went wrong, and how did we catch it?
First, an earlier builder session reported that it had created pages and passed its local checks. But it had never uploaded a build. When the temporary workspace reset, there was no source to recover from Whop. We rebuilt the source and saved it both in our repository and with the uploaded Whop build.
Second, our first hosted draft returned the homepage at the clean offer URLs. The individual HTML files existed, but a visitor opening the Power Hour route still saw the offer hub. Local testing had not exposed that hosting difference.
We fixed the worker’s route mapping and opened all six hosted routes again. Power Hour then displayed Power Hour, Knowledge Panel displayed its own package, and the other pages did the same. That is why I want someone to look at the actual destination. A successful compiler message cannot tell you whether the customer reached the right page.
We also reviewed desktop and mobile layouts, exercised the offer selector, and checked the purchase-link parameters. Our standard for completed work is useful here: evidence must come from where the work landed.
Why are there two different partner paths?
Refer a buyer
An affiliate sends a customer to an LSS offer. That buyer’s affiliate code should travel through the landing page to the relevant checkout. The configured offer commission governs the sale.
Refer a business
A business joins Whop through my business-referral link. This follows Whop’s business-referral terms and is distinct from someone buying one of our services.
We checked that an incoming buyer affiliate code remained on the Power Hour checkout link. We also checked that the business signup link kept my own referral code instead of replacing it with the incoming buyer code.
Preserving those links is a technical check. It does not prove that Whop has credited a referral or paid a commission. I want actual attribution and payout records before claiming revenue. We are also keeping second-tier override rates unpublished until the applicable terms are confirmed in writing.
My Whop guide for businesses explains the broader strategy. For this rollout, the practical next move is one useful implementation that a business owner can inspect and understand.
Does this mean we can cancel Infusionsoft now?
No. We have already moved away from SPP, but publishing landing pages does not complete the replacement of Keap, formerly Infusionsoft. The new pages are one part of the customer journey.
Before retiring the remaining workflows, we still need to prove that a successful payment creates or updates the right contact, applies the correct offer tag, and starts fulfillment exactly once. Refunds, failed payments, cancellations, and completed installment plans need their own checks. Existing customers must not end up with two subscriptions billing at once.
We inspected the Power Hour checkout and its preparation instructions. We did not run a new paid test of the complete payment-to-CRM-to-fulfillment sequence in this build. That is the next operational milestone, and it should stay visible as unfinished work.
How can you use the same approach?
- Start with an offer you already understand. Write down the price, payment schedule, who it serves, what is included, and what happens after purchase.
- Give your existing assistant the source material. Ask it to build from verified offers and real proof. Save the files in a place you control.
- Connect each page to the right action. Use checkout when the offer is ready to buy; use an audit or assessment when scope must be agreed first.
- Check the hosted result. Open each URL, inspect it on a phone-sized screen, and follow its links. Keep the purchase test separate from simply viewing a payment form.
- Document what happened. Record the mistakes, the fixes, and the remaining dependencies so the next implementation starts further ahead.
This is the kind of work I want our AI builders to learn: take a real business problem, implement something useful, and show the evidence. The pages are an example people can visit. The migration work continues from there.
Compare our offersExplore the DIY installation guide
I am a Whop partner and may earn a commission from qualifying businesses referred through my links. Commission eligibility and payment follow the applicable Whop terms. This article reports the build and publication process; it does not claim new sales or affiliate earnings from these pages.