ETHAN+LABFollow along

I GIVE THE DIRECTION. AI DOES THE BUILDING.

Watch me build an entire ecommerce business from scratch by prompting AI.

The art. The brand. The website. The ads. The systems.
GPT-6 is doing the heavy lifting. I’m documenting the whole thing so you can see how far it goes.

Here’s the plan
inkrally.com ACTUAL SITE RECORDING
Homepage → Collection → ProductVisit the store
The real website. Built by AI.A recording of the live store, not a concept mockup.
FROM A PROMPT TO A REAL PRODUCT.Built in public. Follow what happens next.
GET A FRONT-ROW SEAT

Become a watcher.

Enter your email. Join the count. Get the progress, mistakes, and useful lessons as we build this for real.

THE WATCHER COUNTGROWING WITH THE EXPERIMENT
people followingJoin by email. See the count grow.
WAITING FOR THE COUNT

Real active email opt-ins. Counted once per person—not people currently online. Refreshes every 15 seconds and after joining.

AI is building it.
I’m steering it.

I’m Ethan. Instead of hiring a team or building everything myself, I tell GPT-6 what I want, review the result, and give it the next direction.

It researches, designs, writes the code, creates assets with tools, solves problems, and keeps going. My role is judgment, feedback, and the steps that need the business owner.

GPT-6 + ITS TOOLSTHE EXECUTION

Research & strategy

Study the market, products, production requirements, and marketing options.

Brand & artwork

Develop the identity, create art concepts, revise from feedback, and prepare imagery.

Website & code

Build the store, product pages, review tools, integrations, and this public platform.

Marketing & operations

Create stills and video edits, prepare campaigns, document workflows, and work through the systems.

Ethan RogersME, ETHAN

Direction.
Feedback.
Owner decisions.

I prompt, pick favorites, flag what’s wrong, approve the direction, and set up or authorize the accounts.

See both in the log

Here’s what already exists.

REAL OUTPUT, NOT A TO-DO LIST

The brand

Name, wordmark, visual system, and brand guide.

Created

The products

An art catalog, review galleries, and print preparation work.

Created · sample checks ahead

The storefront

A live custom website with collections, product pages, and cart.

Live · order testing ahead

The marketing assets

Photography concepts, still ads, edited videos, and a launch plan.

Prepared · campaigns not launched

There’s a lot behind “AI built it.”

OPEN A WORKSTREAM

This wasn’t one prompt and a template. Here’s the work that went into turning the idea into a brand people could actually buy from.

01Studied the category deeplyMarket & creative research

AI studied Arte Collective and IKONICK in depth: their collections, product presentation, shopping experience, social presence, and publicly visible ads. It also expanded the competitor research beyond those two brands.

Ethan supplied the original reference links. AI carried out the detailed investigation and turned it into recommendations. Public ad visibility informed what to test; competitors’ sales performance was not available.

02Built an identity that could stand aloneBrand development

AI developed the INKRALLY name and visual direction, explored typography and colors, refined the text logo, and created a brand guide with reusable assets and mockups. Ethan rejected the early generic symbol and pushed for a more memorable result.

The output is a consistent brand system that carries through the store, creative, and product presentation.

03Turned taste into better productsArtwork & feedback

AI created art concepts, built a gallery where Ethan could rate every piece and leave notes, then used that feedback to revise favorites, remove weaker ideas, and develop more focused collections.

The important part was the cycle: create, inspect, get specific feedback, revise. A new concept is never automatically a print-approved product.

04Worked through the physical print problemProduction quality

AI investigated canvas requirements, tested larger artwork files, prepared placement checks, and worked through how the front image and stretched sides fit together. An early enlargement changed the art, so the approach had to be corrected.

Digital preparation is done in stages. Physical sample inspection and final production approval remain real checkpoints.

05Built the store and its supporting toolsWebsite & engineering

AI wrote the custom storefront, designed collection and product experiences, built cart interactions, improved mobile layouts, and prepared commerce connections. It also built the private art-review tools and this separate experiment platform.

The site is live. Final checkout-to-fulfillment testing is still ahead; a visible website is not the same as a verified order flow.

06Took the ad creative apart, then rebuilt the approachMarketing & video editing

AI studied competitor creative at the frame and edit level, examined product positioning and changing backgrounds, mapped cuts to music, and produced INKRALLY stills and collection films. Ethan selected the direction and pushed the pacing and finish further.

The research, edits, and workflows are documented. These are creative hypotheses to test, not claims of proven sales. Paid-use clearance and launch checks still come before running campaigns.

THE SIMPLE VERSION

Prompt → Build → Review → Improve.

I describe the next outcome. AI does the work and checks it. I give feedback where judgment matters. We improve it and move to the next real checkpoint.

Don’t just take my word for it.

The actual brand guide, artwork, photography, and creative from this build. Open any image to take a closer look.

PAGES FROM THE ACTUAL BRAND GUIDE
01 / THE BRAND SYSTEM

A real identity.
Down to the details.

AI developed the INKRALLY name and visual direction, explored fonts and color combinations, then carried the chosen system across the logo, website, and creative.

I pushed for a stronger name and brighter branding, and rejected the early generic icon. AI refined the wordmark and packaged the result into a designed brand guide with logo variations, typography, color, and example applications.

WHAT IT PRODUCEDA brand guide, reusable logos, a defined visual system, and packaging and stationery concepts.
Soft Reset
Carbon Mind
Signal Noise
Human Input
Inner Signal
SELECTED ART FROM THE INKRALLY CATALOG
02 / THE ACTUAL ARTWORK

It made the products, too.

AI directed image-generation tools to create the art itself. Then it built a review gallery where I could mark favorites and leave specific notes on every piece.

Those notes became the next round: revise the composition, change the wording, improve a detail, or drop the concept entirely. The stronger directions shaped new collections. AI also worked through enlargement and canvas-edge preparation so an approved image could become a physical product.

WHAT IT PRODUCEDAn original art catalog, repeated concept revisions, and a workflow for selecting and preparing print masters. Physical quality checks still matter.
GENERATED CAMPAIGN IMAGERY · NOT CUSTOMER PHOTOS
03 / PHOTOGRAPHY & PRODUCT PRESENTATION

From a flat image
to a place in a room.

AI created room scenes, canvas mockups, and unboxing imagery to help people picture the art at home. It worked on lighting, materials, scale, and how each artwork appears in the scene.

I called out images that looked artificial. AI regenerated and refined them, then checked product pages so the scene showed the matching artwork. These are generated campaign images; real sample photography is a separate next step.

WHAT IT PRODUCEDInterior imagery, product presentations, unboxing concepts, and an organized photography library.
OUR OWN CREATIVE · PREPARED FOR TESTING
04 / THE MARKETING CREATIVE

It studied the ads.
Then did the editing.

AI studied competitor storefronts, social posts, and public ad-library examples. It broke down how the art was presented, how video cuts landed on beats, and how changing backgrounds kept the product clear.

Then it made INKRALLY stills and collection films, refined the edits from my feedback, and documented the workflow. It also prepared the broader marketing plan. The next test is whether our creative sells—not whether it gets compliments.

WHAT IT PRODUCEDStill ads, beat-synced video variations, organized creative assets, and a staged testing plan. Competitor sales data wasn’t available; campaign launch checks remain ahead.
And the store you saw at the top? AI wrote the code for that, too.Explore INKRALLY

The work, on record.

The decisions, builds, revisions, and human interventions that move this forward.

UPDATED Sep 7, 2026

Prepared a portable offline marketing overview with the existing creative library and 22 playable videos. Verified local links and playback before delivery.

TAKEAWAYUseful working knowledge should travel with its supporting assets.

Published milestone · complete

Added a clearly labeled prelaunch checkout that can privately save an opted-in email and selected artwork. Verified consent, failed-save handling, removal, and the unavailable-payment message on the live site. No orders or charges were enabled.

TAKEAWAYA store can learn from early interest while being clear about what is available and respecting email preferences.

Published milestone · complete

Expanded the storefront to 155 artworks and created five distinct photographic mockups per piece, with updated detail views, sharing images, and collection discovery. Checked product identity and mobile layouts.

TAKEAWAYShowing the exact artwork consistently matters as much as the surrounding presentation. Concept approval remains separate from print production approval.

Published milestone · complete

Approved product-specific photographic scenes, requested a clearer collection-first homepage, and chose to keep payments paused while offering launch reminders.

TAKEAWAYHuman direction sets the commercial priorities; implementation and verification remain separate work.

Published milestone · complete

Updated the private review catalog to reflect the latest selections while retaining original files and feedback.

TAKEAWAYApproval of a concept remains separate from final print production approval.

Published milestone · complete

Marked five new technology concepts as favorites, approved a revised aerospace concept, and rejected two older concepts.

TAKEAWAYSpecific selection keeps the catalog focused.

Published milestone · complete

Captured actual storefront browsing and built a visual walkthrough with selected brand-guide pages, artwork, generated campaign images, and ad designs.

TAKEAWAYShow the real artifacts and explain the work without publishing the private operating playbook.

Published milestone · complete

Requested a real recording of the store, a direct headline, and visible examples of the guide, art, photography, and creative.

TAKEAWAYThe experiment is easier to understand when people can see its outputs.

Published milestone · complete

A reviewed record of major actions, refreshed every minute. Early entries were reconstructed from project history and feedback; their dates are approximate. This is not a complete tool transcript or an independent audit.

The business numbers.

Coming soon: live revenue, orders, ad spend, and store traffic. We’ll connect the actual sources after the commerce setup is ready. A dash means the data is not connected yet.

RevenueComing soonPlanned source: Shopify
OrdersComing soonPlanned source: Shopify
Ad spendComing soonPlanned source: Ad platforms
Store visitsComing soonPlanned source: Web analytics

Revenue is not profit. Future reporting will identify the source, period, and what’s included. Ethan’s existing audience is part of the context.

From building it to selling it.

First make the product and buying experience work. Then test how to sell it. Scale only when the results justify it.

01

Build the business

Create the name, brand, art, store, product imagery, and marketing assets.

WHAT GETS US FORWARDThe digital brand and storefront exist. You can explore them today.
Built
02

Make sure it actually works

Inspect a physical canvas. Finish checkout, fulfillment, order emails, and tracking. Place a real test order.

WHAT GETS US FORWARDAn approved sample and a paid order that reaches fulfillment.
Current checkpoint
03

Start showing the world

Publish the build series, share the art, prepare product listings, and start a small set of marketing tests.

WHAT GETS US FORWARDLearn which art and messages make people visit and consider buying.
Next
04

Find what sells

Compare a few ad ideas and landing pages. Follow the path from a click to a real order. Improve what stops people buying.

WHAT GETS US FORWARDFind repeatable purchases while tracking the full cost of getting each customer.
Then
05

Scale what works

Increase investment gradually in the combinations that earn it. Make more strong creative and keep checking delivery quality.

WHAT GETS US FORWARDGrow only when orders leave money after product, shipping, fees, and advertising.
After proof
06

Show the results

Connect the numbers here. Publish what worked, what failed, and what changed as the business grows.

WHAT GETS US FORWARDA useful public record of revenue, spend, outcomes, and lessons.
Ongoing

Small firsts. Bigger next steps.

These are the goals. We’ll check them off when they happen—and show the work that got us there.

Be part of the first 100
01

The first watcher

Someone chooses to follow the experiment.

COUNT CONNECTING
01

The first real order

A real customer buys a canvas. An owner test order does not count.

PENDING · ORDER DATA COMING SOON
100

The first 100 watchers

One hundred people following along by email.

COUNT CONNECTING
10

The first 10 customers

Ten distinct paying customers, excluding test orders.

PENDING · ORDER DATA COMING SOON
1K

The first 1,000 watchers

A thousand people learning alongside the build.

COUNT CONNECTING
100

The first 100 orders

One hundred paid customer orders, excluding cancellations and tests.

PENDING · ORDER DATA COMING SOON

Watcher goals use the real newsletter count. Customer and order goals will be checked against verified commerce data; they are not claims about current sales. Our scaling goal is to leave money after all variable costs and advertising, not just grow revenue.

NOTE / 01

Taste is a feedback loop.

The useful breakthrough wasn’t generating more art. It was giving precise feedback on what deserved another round.

NOTE / 02

Pixels aren’t a product.

Print resolution, canvas edges, and a physical sample all matter. Looking good on a screen is only the first test.

NOTE / 03

A demo isn’t an operation.

A beautiful store still needs reliable checkout, fulfillment, customer communication, and a real order test.

THE TERMS OF THE EXPERIMENT

Open about the work.
Careful with the details.

The goal is to make this useful to follow, without exposing customers, accounts, or the operating playbook.

What does GPT-6 actually do?

It leads research, writes code, directs creative tools, prepares assets, and documents the work. Image generation and other software are separate tools in that workflow. The log attributes milestones to GPT-6 plus tools, rather than pretending one model does everything alone.

What does Ethan actually do?

I prompt and direct the work, review and select art, make business decisions, set up and own the accounts, provide access privately, and judge quality. Human contributions include feedback and verification—not just the initial prompt.

Is everything happening autonomously?

No. This is an owner-directed experiment with human judgment and approval. The ledger records major actions as they are reviewed and published. It is not surveillance of every click or a claim that the business runs unattended.

What will be public?

The broad roadmap, meaningful milestones, revisions, lessons, and eventually clearly sourced business results. Customer information, credentials, exact campaign operations, and proprietary production instructions stay private.

Is this starting with no audience?

No. I already make content and have an audience. The series may bring traffic to the store. That advantage is part of the experiment, and future reporting should distinguish content-driven traffic from paid acquisition.

Is this an official OpenAI project?

No. This is an independent experiment by Ethan Rogers. It is not affiliated with or sponsored by OpenAI. There is no guarantee of revenue, profit, or a particular outcome.