Product Ad Studio: Run a Full Ad Campaign in One Claude Conversation

You provide a brief and reference assets. Claude plans the shots, generates images through OpenArt MCP, adds copy as an overlay, and prepares the files for download, all in one conversation.

The problem with a normal product ad campaign

A normal product ad campaign slows down at the handoffs. You send a brief and product references to a photographer, wait for the shoot, pass selected images to a designer, and send layouts to a copywriter. The campaign moves through at least three tools, with people and file names carrying context between them.

Each revision can restart part of that chain. A new headline sends the layout back to design. A different crop may expose a weak composition and require another image. Feedback spreads across chat threads, documents, and exported files, so even a small campaign can take a week of back and forth.

A better prompt cannot remove those handoffs. You need one conversation that can read the brief, inspect the product assets, plan the required shots, generate options, and apply copy after you choose the winners. The conversation keeps the campaign context in one place, while you make creative decisions instead of translating them between specialists and tools.

Connecting OpenArt to Claude

1. In Claude, click your profile icon and open Settings.
2. Select Connectors.
3. Browse the available connectors or search for OpenArt.
4. Select OpenArt and click Connect.
5. When the authorization window opens, allow Claude to access your OpenArt account.

Once access is approved, return to Claude. OpenArt should now appear among your connected services. You can follow the same connector setup in ChatGPT if you want to run the Product Ad Studio workflow there instead.

Installing the Product Ad Studio skill

Open Claude’s Settings, then follow Customize → Skills → Add → Upload skill. Drag the Product Ad Studio skill file into the upload area and complete the installation.

Start the workflow by telling Claude, “I want to create a product ad.” You do not need a specialized prompt. If you have no brief, Claude asks questions about your product, campaign goals, formats, and creative direction. If you already have a brief, upload it with reference assets such as product photos and a logo. Claude reads those materials and uses them to plan the campaign.

Feeding Claude a sample campaign brief

You start the Belch beer campaign with two kinds of input. The one-page brief specifies the required ad formats and visual mood. It also sets the color palette and tone, then supplies several headline options. Alongside that document, you upload the Belch logo and reference photos of the bottle and can. The brief tells Claude what the campaign should communicate, while the images show the exact packaging it must preserve.

Claude reads those inputs as production instructions rather than waiting for a separate prompt for every image. It maps each requested format to the relevant product and proposes two formats for the can and two for the bottle. Each planned shot carries forward the brief’s mood, colors, copy direction, and product reference.

You review that shot plan before Claude generates anything. If a format or product treatment looks wrong, you can correct the plan in the same conversation instead of discovering the mismatch after spending credits. Once the plan matches the brief, Claude has a defined set of deliverables ready for image generation.

Generating the shots with Nano Banana 2

Claude turns the approved shot plan into a controlled Nano Banana 2 generation run. Product Ad Studio defaults to four base images for each shot, giving you several compositions to compare before choosing a direction. With four planned shots, that default produces 16 base images across the can and bottle formats.

Before Claude starts generating, you can review the quantity and cost. You can reduce the number of base images to conserve credits or increase it when you want more options. Claude updates the cost before you commit, so the AI product ad generator cannot quietly run up a larger batch than you intended.

Once you approve the settings, OpenArt generates the planned shots through Claude. Each output follows the selected product, format, and visual direction while giving you enough variation to judge framing and product presentation. The next decision stays with you. You choose which images move forward rather than accepting the first result.

Picking winners and adding copy after the fact

Once Nano Banana 2 returns the options, you choose the strongest can and bottle shots by copying each image and pasting it back into the Claude chat. Claude treats those pasted images as your selections, so you hand-pick the winners instead of asking it to guess which ones you prefer. Repeat the step until each required format has a selected base image.

Add the headline and supporting copy after you lock those images. Baking the Belch campaign text into every generation would force you to regenerate an otherwise usable product shot whenever the wording changes. Text overlays let you test headline options from the brief while preserving the selected image.

For example, you can keep one winning can shot and try several headlines over it. Each copy revision takes less time and avoids the cost of another image generation.

From four images to a downloadable campaign

Once you approve the four final images and their text overlays, the campaign is ready to export. You can download each ad individually for a specific placement, or download the full set for handoff and publishing.

The Belch beer campaign reaches that point within minutes. Claude keeps the brief, selected shots, and copy decisions in one conversation, so you avoid days of file passing between a photographer, designer, and copywriter. You finish with a consistent set of AI generated product ads instead of disconnected drafts.

Where this goes next: turning the hero shot into video

The same Claude conversation can carry the Belch campaign into video. You can reuse the brief, logo, can, bottle, and selected hero image instead of rebuilding the creative direction for a separate tool.

The follow-up will show how to turn that hero shot into a moving product spot through OpenArt MCP. Your still campaign becomes the source material for motion, so both formats come from the same assets and decisions.

FAQs

What happens if you do not have a campaign brief?

A campaign brief gives Claude the product, formats, visual direction, and copy requirements. Without one, OpenArt’s Product Ad Studio skill asks questions and builds the plan with you. You can start with a rough product idea instead of preparing a formal document.

Does Product Ad Studio work in ChatGPT too?

Product Ad Studio can run in ChatGPT through the OpenArt MCP connector. You use the same skill and provide the same brief or reference assets. See how to connect OpenArt MCP to ChatGPT if you want to set that up. You can choose the assistant that already holds your campaign context.

What is Nano Banana 2?

Nano Banana 2 is the image model used to generate the base product shots in this walkthrough. OpenArt runs the model after Claude plans the shots and shows the generation cost. You can review the quantity and cost before committing credits.

Can you use the generated ads commercially?

OpenArt gives you ownership of your generated content, with commercial use rights on eligible plans. Commercial rights currently begin with the Plus plan. You can use qualifying outputs in ads, client work, merchandise, and social posts without royalties or attribution requirements.