Visual Insights for AI: The Feng-GUI MCP Server

Visual Insights for AI: The Feng-GUI MCP Server

Digital agencies and research agencies are constantly looking for ways to back their creative strategies with objective data. While AI assistants have made generating and refining ideas faster, they often lack a structured way to evaluate how humans actually perceive visual media. 

To bridge this gap, we are excited to launch the Feng-GUI MCP Server in beta. By wrapping the Feng-GUI API into a Model Context Protocol (MCP) server, we are making it possible for your AI agents to access visual insights natively within their workflows.

Bring Visual Analysis into AI Workflows

The Model Context Protocol (MCP) allows AI models, such as Claude, to connect securely and dynamically with external APIs. For agencies and research teams, this means your AI workflows can now call upon Feng-GUI directly to:

• Predict attention: See where users are most likely to focus within the first few seconds of viewing an asset.
• Analyze emotional responses: Gain a deeper understanding of how visual elements might emotionally resonate with viewers.
• Measure memorability: Evaluate which parts of an image, campaign, or interface are most likely to be remembered.

Your AI assistants can then process these visual insights in real-time to suggest precise, data-driven improvements to layouts, advertisements, or packaging designs.

Empowering Designers with Smarter Iteration

UX, UI, and graphic designers can also benefit immensely from this integration. Instead of relying on manual testing cycles, designers can ask their AI assistants to evaluate layouts, review attention scores, and suggest actionable adjustments. This helps ensure that key visual elements, such as Call-to-Action (CTA) buttons or brand logos, stand out as intended before a single mock-up is finalized.

Access Details and Setup

The Feng-GUI MCP server is currently in beta. To utilize this integration, you must be a registered Feng-GUI customer with the 'api' role enabled on your account. 

If you meet these requirements, you can get started by connecting to the MCP server via Server-Sent Events (SSE) at: https://mcp.feng-gui.com/sse

Feng-GUI API documentation https://docs.api.feng-gui.com

We invite agencies, researchers, and designers to explore this integration and bring empirical visual insights to their daily AI workflows.

Updated DPA, Sub-processors, and AI Terms

At Feng-GUI, we know that as AI becomes a larger part of your daily workflow, data privacy and compliance are more important than ever. Whether you are a small agency or a Fortune 500 brand, you need total peace of mind when uploading unreleased designs and ad campaigns.

To meet strict enterprise standards and global privacy regulations (like GDPR and the upcoming EU AI Act), we have officially published our Data Processing Agreement (DPA) and completely updated our AI Terms.

Here is a simple summary of our new updates:

New DPA & Sub-processor Transparency
We have published a formal Data Processing Agreement (DPA). This legal document outlines exactly how we handle and protect your data. Alongside it, we have listed our approved sub-processors, the secure, third-party infrastructure services we use to run Feng-GUI. You now have 100% transparency into where and how your data is processed.

No AI Model Training
We have made our legal language crystal clear: we explicitly guarantee that your private inputs and outputs are never used to train, test, or fine-tune our core AI models or any third-party models. Your data belongs only to you.

Your Data is Isolated (Logical Separation)
We added explicit guarantees that your data is completely isolated. Your inputs (like images) and outputs (like heatmaps) are kept strictly separate from other customers, third-party providers, and our own internal data. We never mix your proprietary designs into a shared "data lake."

Enterprise Compliance (Auditable Logs)
To support the strict compliance requirements of our enterprise clients, we now formally maintain secure audit logs of AI feature usage. If your company's IT or legal team ever needs to verify what data went into the AI and what came out for an internal audit, we have the records ready for you.

How Our Predictive AI Works
We added a few standard industry clarifications about how predictive AI operates. Because our AI predicts human attention, if two users upload similar designs, the AI may generate similar insights. We also included a quick reminder that while our platform provides highly accurate mathematical predictions of human behavior, they are still predictions!

What this means for you
These updates do not change how you use Feng-GUI or affect your pricing. They simply provide you and your legal teams with the concrete, contractual guarantees needed to use our tools safely and confidently.

You can review the full updated agreements here:

DPA & Sub-processors: https://feng-gui.com/dpa

AI Terms: https://feng-gui.com/ai

As always, if you or your legal team have any questions, feel free to reach out to us!

New Aesthetics Report - Memory Heatmap

Introducing the New Aesthetics Report

We’ve redesigned our Aesthetics Report to make it more intuitive and actionable.

The new report (with Memory heatmap) will go live during February 2026.

The old report (with Balance lines) is deprecated and will be obsolete on September 1st 2026.

The previous report focused on vertical and horizontal balance metrics. While technically accurate, these indicators were often hard to interpret and didn’t clearly explain what to change in an image.

The updated report now centers on visual memorability, combining a clear Memory Score with an easy-to-read Memory Heatmap.

The heatmap shows how attention is distributed across the image:

• Red – Highly noticeable and memorable areas
• Yellow – Medium memorability
• Blue – Low memorability

Instead of abstract balance values, the new report shows what viewers actually will remember. This makes it easier to validate focal points, reduce visual competition, and optimize layouts with confidence.

The result is a more human-centered, actionable aesthetic analysis designed to support real creative decisions.

Postman Collection and Examples of Feng-GUI API

Postman Collection

For practical examples of request syntax and hands-on testing, you may find the API documentation and collection available on Postman helpful. 

These resources are commonly used for learning, exploration, and validating API requests alongside this documentation

https://docs.api.feng-gui.com/

GET (URL)
In addition to the standard POST syntax, for simplified access or quick testing via a browser, the API also supports requests via the GET method. In this format, the method name is appended to the base URL as a path, and parameters are passed as standard URL query strings.

GET URL Structure:
https://service.feng-gui.com/json/api.ashx/[METHOD]?id=[ID]&[PARAM1]=[VALUE1]&[PARAM2]=[VALUE2]

Example GET Request:

https://service.feng-gui.com/json/api.ashx/ImageAttention?id=1&InputImage=https://service.feng-gui.com/users/USERNAME/files/images/IMAGEID.png&ViewType=0&ViewDistance=0&AnalysisOptions=0&OutputOptions=0

Demo page of GET API methods
FG-JSON-GET-Demo.html

for more, see the main API page at https://feng-gui.com/developers