AI Image Generation Action Now Accepts Reference Images

TL;DR

The AI Image Generation action in workflows now supports reference images. This is a small update for agencies using AI image generation.

Who this affects

HighLevel has not confirmed who this specifically affects. Agencies utilizing the AI Image Generation action within their workflows for client services or internal processes will be interested. Agencies not currently using AI image generation will find this update largely irrelevant.

What changed

The AI Image Generation action now accepts reference images. Users can attach up to five reference images to a single AI Image Generation action. These reference images can be added in three ways: uploading from a system, picking from the Media Library, or providing a URL. The URL field specifically supports custom values, which can come from webhook payloads, contact fields, or previous actions within the workflow. Images are passed to the AI model in the order they are added. Prompts can address each image by its positional order.

HighLevel has not confirmed any limits or plan gating for this feature. HighLevel has not confirmed the rollout status for this feature.

Where to find it

Agencies can find this new capability by opening the AI Image Generation action within a workflow. Scroll down to the “Reference images” section, which is located below the templates.

How agencies will use this

An agency running a marketing campaign for a real estate client might use this. The client wants social media posts featuring specific house styles. Instead of just describing the house, the agency can set up a workflow. This workflow uses a trigger, perhaps a form submission or a custom field update, to initiate the AI Image Generation action. Inside the action, the agency uploads or links to reference images of the exact architectural styles – a Tudor home, a modern minimalist design, a Victorian house – up to five images. The prompt then instructs the AI model to generate new images in a similar style, referencing "image 1" for Tudor, "image 2" for modern, and so on. This ensures brand consistency and adherence to client visual guidelines across automatically generated content. The output image can then be used in an email or social media post action.

Our take

This is a standard update. It provides more control and precision when using HighLevel's AI Image Generation. It does not fundamentally change how agencies operate or onboard clients. Agencies already leveraging AI for content creation will find it useful for improving output accuracy. For others, it's a minor enhancement in a niche feature.