# Workflow Action – AI Analyze Image > Learn how to use AI Analyze Image in GHL Customer Care workflows to analyze images, extract text, configure prompts, use outputs, and automate visual data tasks. - Source: https://docs.ghlcustomercare.com/docs/workflows/workflow-ai-workflow-actions/workflow-action-ai-analyze-image - Section: Workflows / Workflow AI Workflow Actions - Reading time: 9 min - Images: 8, each explained below --- This article explains how to use the AI Analyze Image action in workflows to analyze images with AI vision. You’ll find step-by-step configuration instructions, image URL requirements, how to use the output, practical use cases, and answers to common questions. **Note:** This is a **premium** **action**. Using this action will incur additional charges per execution. --- **TABLE OF CONTENTS** - [What is AI Analyze Image Action in Workflows?](#what-is-ai-analyze-image-action-in-workflows)[](#key-benefits-of-ai-analyze-image-action) - [Key Benefits of AI Analyze Image Action](#key-benefits-of-ai-analyze-image-action)[](#action-details) - [Action Details](#action-details)[](#how-to-use-the-ai-analyze-image-action) - [How to Use the AI Analyze Image Action](#how-to-use-the-ai-analyze-image-action)[](#image-url-requirements) - [Image URL Requirements](#image-url-requirements)[](#using-the-output-in-downstream-actions) - [Using the Output in Downstream Actions](#using-the-output-in-downstream-actions)[](#use-cases) - [Use Cases](#use-cases)[](#tips-for-better-results) - [Tips for Better Results](#tips-for-better-results)[](#pricing) - [Pricing](#pricing)[](#frequently-asked-questions) - [Frequently Asked Questions](#frequently-asked-questions)[](#related-articles) - [Related Articles](#related-articles) --- # **What is AI Analyze Image Action in Workflows?** The AI Analyze Image action, available under AI Actions in Workflows, uses AI vision models to analyze an image and return a text response. You provide the image URL and a prompt describing what you want the AI to look for, and the action returns its analysis as text you can use in downstream actions. This is useful for extracting text from images (OCR), generating image descriptions, identifying objects or attributes, and verifying image content – giving your workflows the ability to interpret visual input rather than just pass it along. --- ## **Key Benefits of AI Analyze Image Action** - **Text Extraction (OCR):** Pull text out of screenshots, receipts, documents, and other image-based content so it can be used as workflow data. - **Image Descriptions:** Generate detailed descriptions of what an image contains, useful for cataloguing, alt text, or content summaries. - **Flexible Prompting:** Ask the AI anything about the image – the prompt controls exactly what it looks for and how it responds. - **Adjustable Detail Level:** Choose how thoroughly the image is analyzed, balancing depth against speed. - **Pairs with Image Generation:** Combine with the AI Image Generation action to describe or verify an image your workflow just created. --- ## **Action Details**

Field	Description
Action Name	A custom name for the action, used to identify it on the workflow canvas and in execution logs.
Model	The AI vision model used for analysis. Two models are available: GPT-5.6 Luna and GPT-5.6 Tera.
Image URL	The URL of the image to analyze. Supply it from a previous step or from a custom value. The URL must be publicly accessible – see Image URL Requirements below.
Analysis Prompt	Instructions telling the AI what to look for in the image and how to respond. Supports custom values for dynamic prompts.
Detail Level	How thoroughly the image is analyzed. Defaults to Auto; can be set to High or Low.

--- ## **How to Use the AI** **Analyze** **Image Action** ### **Step 1:** Add the Action 1. Go to **Automation** > **Workflows**. 2. **Create** a new **workflow** or **edit** an **existing** one. ![Step 2 of 3: Create a new workflow or edit an existing one](https://jumpshare.com/share/sdAZAU36nE7nJWCjV65F+/Screen+Shot+2026-08-26+at+16.44.40.png) **Step 2 of 3: Create a new workflow or edit an existing one** This screenshot accompanies step 2 of 3 in the "Step 1: Add the Action" section of "Workflow Action – AI Analyze Image". At this point in the walkthrough you create a new workflow or edit an existing one. The navigation path used here is Automation > Workflows. The next step is to in your workflow, click the + icon to add a new action, then find AI Analyze Image under AI Actions and select it. Give the action a clear, descriptive name so it’s easy to identify on the canvas. - Path: Automation > Workflows - What this covers: Create a new workflow or edit an existing one. - Controls: Create, workflow, edit, existing, Automation, Workflows, +, add, new, action, AI Analyze Image, under AI Actions - Next: In your workflow, click the + icon to add a new action, then find AI Analyze Image under AI Actions and select it. Give the action a clear, descriptive name so it’s easy to identify on the canvas. Full procedure: 1. Go to Automation > Workflows. 2. Create a new workflow or edit an existing one. <- this image 3. In your workflow, click the + icon to add a new action, then find AI Analyze Image under AI Actions and select it. Give the action a clear, descriptive name so it’s easy to identify on the canvas. 3. In your workflow, click the **+** icon to **add** a **new** **action**, then find **AI Analyze Image** **under AI Actions** and select it. Give the action a clear, descriptive name so it’s easy to identify on the canvas. ![Step 3 of 3: In your workflow, click the + icon to add a new action, then find AI…](https://jumpshare.com/share/kQLCL5FaVCUr4s1hKvvY+/GIF+Recording+2026-08-28+at+17.26.42.gif) **Step 3 of 3: In your workflow, click the + icon to add a new action, then find AI…** This animation accompanies step 3 of 3 in the "Step 1: Add the Action" section of "Workflow Action – AI Analyze Image". At this point in the walkthrough you in your workflow, click the + icon to add a new action, then find AI Analyze Image under AI Actions and select it. Give the action a clear, descriptive name so it’s easy to identify on the canvas. The navigation path used here is Automation > Workflows. - Path: Automation > Workflows - What to choose: In your workflow, click the + icon to add a new action, then find AI Analyze Image under AI Actions and select it. Give the action a clear, descriptive name so it’s easy to identify on the canvas. - Controls: +, add, new, action, AI Analyze Image, under AI Actions, Automation, Workflows, Create, workflow, edit, existing Full procedure: 1. Go to Automation > Workflows. 2. Create a new workflow or edit an existing one. 3. In your workflow, click the + icon to add a new action, then find AI Analyze Image under AI Actions and select it. Give the action a clear, descriptive name so it’s easy to identify on the canvas. <- this image ### **Step 2:** Select the Model Choose the AI vision model to use for analysis. Two models are available: - GPT-5.6 Luna - GPT-5.6 Tera If you’re unsure which to use, run the same image and prompt through both and compare the responses before publishing the workflow. ![Step 2: Select the Model (image 3 of 8)](https://jumpshare.com/share/U2OY2Fou3rn6EuItyWye+/Screenshot+2026-08-28+at+17.34.07.png) **Step 2: Select the Model (image 3 of 8)** This screenshot appears in the "Step 2: Select the Model" section of "Workflow Action – AI Analyze Image". The text alongside this image reads: If you’re unsure which to use, run the same image and prompt through both and compare the responses before publishing the workflow. This part of the guide covers 2 fields, listed below. Immediately after, the guide continues: Enter the URL of the image you want to analyze. - What this covers: If you’re unsure which to use, run the same image and prompt through both and compare the responses before publishing the workflow. - Fields: GPT-5.6 Luna, GPT-5.6 Tera - Next: Enter the URL of the image you want to analyze. Full procedure: 1. GPT-5.6 Luna 2. GPT-5.6 Tera ### **Step 3:** Provide the Image URL **Enter** the **URL** of the **image** you want **to** **analyze**. You can pull this from a previous step in the workflow – for example, the Image URL output of an [AI Image Generation action](https://docs.ghlcustomercare.com/docs/workflows/workflow-ai-workflow-actions/workflow-action-ai-image-generation) – or from any custom value that holds an image URL. The **URL must be publicly accessible.** ![Step 3: Provide the Image URL (image 4 of 8)](https://jumpshare.com/share/Z8S6ZdH2R7fdMm7vr2Ht+/GIF+Recording+2026-08-28+at+17.48.19.gif) **Step 3: Provide the Image URL (image 4 of 8)** This animation appears in the "Step 3: Provide the Image URL" section of "Workflow Action – AI Analyze Image". The text alongside this image reads: You can pull this from a previous step in the workflow – for example, the Image URL output of an AI Image Generation action – or from any custom value that holds an image URL. The URL must be publicly accessible. Immediately after, the guide continues: Write a prompt telling the AI what to look for in the image and how to respond. The more specific your prompt, the more useful the output. - What this covers: You can pull this from a previous step in the workflow – for example, the Image URL output of an AI Image Generation action – or from any custom value that holds an image URL. The URL must be publicly accessible. - Controls: Enter, URL, image, to, analyze, URL must be publicly accessible. - Next: Write a prompt telling the AI what to look for in the image and how to respond. The more specific your prompt, the more useful the output. ### **Step 4:** Enter the Analysis Prompt Write a prompt telling the AI what to look for in the image and how to respond. The more specific your prompt, the more useful the output. **Examples:** - Describe this product image in detail. - Extract all text visible in this image. - List the items visible in this photo. - Read the total amount from this receipt and return only the number. The prompt field supports custom values, so you can build the prompt dynamically from contact fields, web-hook data, or earlier action output. ![Step 4: Enter the Analysis Prompt (image 5 of 8)](https://jumpshare.com/share/BrqlEWMFrFuZdOoyfJrw+/Screen+Shot+2026-08-28+at+17.54.02.png) **Step 4: Enter the Analysis Prompt (image 5 of 8)** This screenshot appears in the "Step 4: Enter the Analysis Prompt" section of "Workflow Action – AI Analyze Image". The text alongside this image reads: The prompt field supports custom values, so you can build the prompt dynamically from contact fields, web-hook data, or earlier action output. Immediately after, the guide continues: Choose how thoroughly the image should be analyzed. - What this covers: The prompt field supports custom values, so you can build the prompt dynamically from contact fields, web-hook data, or earlier action output. - Next: Choose how thoroughly the image should be analyzed. Full procedure: 1. Describe this product image in detail. 2. Extract all text visible in this image. 3. List the items visible in this photo. 4. Read the total amount from this receipt and return only the number. ### **Step 5:** Set the Detail Level **Choose** **how** **thoroughly** the **image** **should** **be** **analyzed**. The **default** is **Auto**, which lets the system pick an appropriate level. You can also select **High** for more careful analysis of fine detail, or **Low** for a faster, lighter pass when you only need a broad read of the image. ![Step 5: Set the Detail Level (image 6 of 8)](https://jumpshare.com/share/NZGbmcfO6bnZOzWBZtoU+/Screenshot+2026-08-28+at+18.01.36.png) **Step 5: Set the Detail Level (image 6 of 8)** This screenshot appears in the "Step 5: Set the Detail Level" section of "Workflow Action – AI Analyze Image". The text alongside this image reads: The default is Auto, which lets the system pick an appropriate level. You can also select High for more careful analysis of fine detail, or Low for a faster, lighter pass when you only need a broad read of the image. Immediately after, the guide continues: Save the action. The analysis output will be available as a variable in all downstream actions once the workflow executes. - What to choose: The default is Auto, which lets the system pick an appropriate level. You can also select High for more careful analysis of fine detail, or Low for a faster, lighter pass when you only need a broad read of the image. - Controls: Choose, how, thoroughly, image, should, be, analyzed, default, Auto, High, Low - Next: Save the action. The analysis output will be available as a variable in all downstream actions once the workflow executes. ### **Step 6:** Save the Action **Save** the **action**. The analysis output will be available as a variable in all downstream actions once the workflow executes. ![Step 6: Save the Action (image 7 of 8)](https://jumpshare.com/share/b5dO4SBiQDmZxj1Kcu7t+/Screen+Shot+2026-08-28+at+18.05.31.png) **Step 6: Save the Action (image 7 of 8)** This screenshot appears in the "Step 6: Save the Action" section of "Workflow Action – AI Analyze Image". The text alongside this image reads: Save the action. The analysis output will be available as a variable in all downstream actions once the workflow executes. Immediately after, the guide continues: The image URL you provide must be publicly accessible. The AI model fetches the image directly from the URL, so anything that blocks anonymous access will cause the action to fail. - How to confirm: Save the action. The analysis output will be available as a variable in all downstream actions once the workflow executes. - Controls: Save, action - Next: The image URL you provide must be publicly accessible. The AI model fetches the image directly from the URL, so anything that blocks anonymous access will cause the action to fail. --- ## **Image URL Requirements** The image URL you provide must be publicly accessible. The AI model fetches the image directly from the URL, so anything that blocks anonymous access will cause the action to fail. **Tip:** A quick way to check is to open the URL in a private or incognito browser window. If the image loads without signing in, the action can reach it. Images stored in Media Storage – including those created by the AI Image Generation action – are already hosted at publicly accessible URLs and work without any extra setup. **This means the URL must not:** - Require a login, password, or authentication token to view. - Sit behind a private share link, permission-restricted drive, or internal network. - Point to a page containing the image rather than the image file itself. --- ## **Using the Output in Downstream Actions** Once the action executes, its response is available in any subsequent action through the custom value picker under AI Analyze Image. The **output is a text value containing the AI’s answer to your prompt** – a description, extracted text, a list, or whatever else your prompt asked for. Because the output is plain text, you can use it anywhere a text value is accepted: save it to a contact custom field, include it in an email or SMS, pass it to an outbound webhook, or feed it into another AI action for further processing. ![Using the Output in Downstream Actions (image 8 of 8)](https://jumpshare.com/share/iUgQiDneO23xj0y0qbfR+/GIF+Recording+2026-08-28+at+18.16.48.gif) **Using the Output in Downstream Actions (image 8 of 8)** This animation appears in the "Using the Output in Downstream Actions" section of "Workflow Action – AI Analyze Image". The text alongside this image reads: Because the output is plain text, you can use it anywhere a text value is accepted: save it to a contact custom field, include it in an email or SMS, pass it to an outbound webhook, or feed it into another AI action for further processing. Immediately after, the guide continues: Scenario: Customers send photos of receipts, invoices, or documents over MMS. You want to read the text out of those images and store it against the contact record instead of reviewing each one manually. - How to confirm: Because the output is plain text, you can use it anywhere a text value is accepted: save it to a contact custom field, include it in an email or SMS, pass it to an outbound webhook, or feed it into another AI action for further processing. - Next: Scenario: Customers send photos of receipts, invoices, or documents over MMS. You want to read the text out of those images and store it against the contact record instead of reviewing each one manually. --- ## **Use Cases** ### **1\. Extracting Text from Customer-Submitted Images** **Scenario:** Customers send photos of receipts, invoices, or documents over MMS. You want to read the text out of those images and store it against the contact record instead of reviewing each one manually. **Setup:** - **Trigger:** Customer Replied (with image attachment) - **Action 1:** AI Analyze Image – Image URL from the attachment. Prompt: “Extract all text visible in this image.” Detail Level: High. - **Action 2:** Update Contact – Save the extracted text to a custom field for review or follow-up. ## **2\. Verifying a Generated Image** **Scenario:** Your workflow generates an image with the AI Image Generation action, and you want a second check on what was actually produced before it goes out to a customer. **Setup:** - **Action 1:** AI Image Generation – Generate the image from your prompt. - **Action 2:** AI Analyze Image – Use the Image URL output from the previous action. Prompt: “Describe this image in detail.” - **Action 3:** If/Else or Internal Notification – Route the workflow based on the description, or send it to your team for a quick sanity check before delivery. ### **3\. Describing Product Images from a Web-hook** **Scenario:** Product images arrive from an external system with no accompanying copy. You want a written description generated automatically for each one. **Setup:** - **Trigger:** Inbound Web-hook (carrying the image URL) - **Action 1:** AI Analyze Image – Prompt: “Describe this product image in detail, including colour, material, and style.” - **Action 2:** Outbound Web-hook – Send the description back to your catalogue or content system. **More Ideas** - **Screenshot triage:** Read error messages out of support screenshots so tickets can be routed by content rather than by hand. - **Alt text generation:** Produce accessibility descriptions for images used across your sites and campaigns. - **Image tagging:** Ask the model to return a short list of categories or keywords, then use that output to tag contacts or route the workflow. - **Submission checks:** Confirm an uploaded image contains what was asked for before it moves further through the process. --- ## **Tips for Better Results** - Be specific about the output you want. “Return only the invoice number” gives you a usable value; “Tell me about this image” gives you a paragraph you’ll have to parse. - Use High detail for small text, fine print, and dense documents. Auto is fine for general descriptions. - Verify the image URL is publicly reachable before publishing – this is the most common cause of the action failing. - If you plan to store the output in a custom field, ask the prompt to return the value in a consistent format so downstream steps can rely on it. - Test with a few real images before running the workflow at volume – image quality varies more than text input does. --- ## **Pricing** Image analysis is billed per execution. The figures below give you an idea of what analyzing an image normally costs. Most analyses fall within this range. **Note:** Treat these figures as a planning estimate rather than a fixed rate. | Metric | Cost per analysis | In cents | | --- | --: | --: | | Median | $0.0004 | 0.04¢ | | Average | $0.0013 | 0.13¢ | Your actual cost per analysis will vary depending on how you configure the action. The main factors are: - **Model** – the vision model you select. - **Detail Level** – the level you set (Auto, High, or Low). High analyzes the image more thoroughly and costs more than Low. --- ## **Frequently Asked Questions** **Q: How much does it cost to analyze an image?** Analyzing an image normally costs between $0.0004 (0.04¢) and $0.0013 (0.13¢). Your actual cost depends on the model and the Detail Level you select. See the Pricing section above for more detail. **Q: What happens if the image URL is behind authentication?** The action will not work. The AI model fetches the image directly from the URL, so it must be publicly accessible without a login, token, or permission check. Open the URL in a private browser window to confirm it loads for anyone. **Q: Where do I get the image URL from?** From any previous step in the workflow that produces one – such as the Image URL output of an AI Image Generation action, an inbound webhook payload, or a message attachment – or from a custom value where you’ve stored an image URL. **Q: Which model should I choose?** Two vision models are available: GPT-5.6 Luna and GPT-5.6 Tera. If you’re unsure, run the same image and prompt through both and compare the responses before publishing the workflow. **Q: What does the Detail Level setting affect?** It controls how thoroughly the image is analyzed. Auto (the default) lets the system decide. High examines the image more carefully, which helps with small text and fine detail. Low is a faster, lighter pass suited to broad descriptions. **Q: Can I use this to extract text from images (OCR)?** Yes. Set your prompt to something like “Extract all text visible in this image” and the action returns the text as its output. For receipts, invoices, and other small or dense text, set the Detail Level to High. **Q: What format is the output in?** The output is a text value containing the AI’s response to your prompt. It’s available in downstream actions through the custom value picker under AI Analyze Image, and can be used anywhere a text value is accepted. **Q: Can I use this together with the AI Image Generation action?** Yes. Add AI Analyze Image after an AI Image Generation step and pass in the generated Image URL. This lets you describe or verify what was produced before the image is sent onward. Images created by AI Image Generation are stored at publicly accessible URLs, so no extra setup is needed. **Q: Can I analyze more than one image in a single action?** Each AI Analyze Image action processes one image URL. To analyze several images in a workflow, add a separate action for each, and reference each action’s output individually in your downstream steps. --- ### **Related Articles** - [Workflow Action – AI Image Generation](https://docs.ghlcustomercare.com/docs/workflows/workflow-ai-workflow-actions/workflow-action-ai-image-generation) [ ](https://docs.ghlcustomercare.com/docs/workflows/workflow-ai-workflow-actions/workflow-action-ai-extract-data) - [Workflow Action - AI Extract Data](https://docs.ghlcustomercare.com/docs/workflows/workflow-ai-workflow-actions/workflow-action-ai-extract-data) - [Workflow Action - AI Translate](https://docs.ghlcustomercare.com/docs/workflows/workflow-ai-workflow-actions/workflow-action-ai-translate) - [Workflow Action - AI Agent](https://docs.ghlcustomercare.com/docs/workflows/workflow-ai-workflow-actions/workflow-action-ai-agent) --- Documentation for GHL Customer Care. Support: support@ghlcustomercare.com