🎥 Week 32 - Turn a product photo into a launch-ready social post

Spend 15 minutes with Andre building an agent that turns a single product photo into a polished, on-brand social post. You give it the photo and the campaign vibe, and it does the rest, generating the graphic and writing the copy ready to publish.

Andre uses a sneaker launch as the live example, so you can see exactly how it handles a real campaign from brief to finished post, then put it to work for your own launches.

Want to build this agent yourself?

Here's the exact prompt Andre uses in the session.

Copy it straight into Autohive:

I need an AI agent that takes a product photo and a short creative brief, then handles everything needed to create and publish an Instagram post, no manual steps required.

What the agent needs to do

Collect three things from the user before starting:

A product photo

A short brief describing the mood or angle for the post (e.g. "bold colourway, limited drop, high energy")

A webhook URL to send the finished post to


Write an Instagram caption based on the brief:

Maximum 2 sentences, punchy and scroll-stopping

Opens with a hook

Ends with relevant hashtags

Matches the tone and energy of the brief


Generate a campaign image:

Before generating, check a folder called references in the content area for any brand style guides, mood boards, or visual guidelines, and use them to inform the image

Use the product photo, the brief, and those brand references to write and execute an image generation prompt


Upload the generated image to Google Drive:

Name the file clearly, for example social-post-brief-summary-date.png

Record the file ID returned after upload


Make the file shareable:

Set permissions so anyone with the link can view it

Record the download link in the metadata of the file. It must be the download link, not the webview link


Fire a webhook to the URL the user provided, with this exact payload:

The Google Drive file ID

The shareable download link

The caption

Confirm once the webhook has fired, keep it brief, one or two lines max


How it should behave

Efficient and direct, no over-explaining

If the user provides all three inputs upfront, skip straight to generating

If the brief is vague, ask one focused clarifying question before proceeding

The agent should check the caption with the user before publishing


Integrations needed

Google Drive, to upload the image and share it

API Call, to fire the publish webhook

Image generation, to create the campaign image (Nano Banana Pro)

Try Autohive free: https://autohive.com