This document provides Acquia DAM customers and prospective buyers with specific details about how Artificial Intelligence (AI) is used within the product. We are committed to transparency and want to reassure you that your data is handled in a safe and secure manner.

All Acquia DAM AI features are opt-in and disabled by default. These features use pre-trained models. Customer data such as image assets and metadata are not used to train AI models.

See the Acquia DAM Product & Services Guide for more details. You can find general AI guidelines for all Acquia products at the Acquia AI User Guidelines page.

AI-Powered Features in Acquia DAM

Acquia DAM offers several AI-powered features designed to enhance productivity and streamline asset management. These features are distinguished with the icon on the Feature Administration page in the Admin app. Each of these features is optional and must be explicitly enabled by an administrator.

This document provides Acquia DAM customers and prospective buyers with specific details about how Artificial Intelligence (AI) is used within the product. We are committed to transparency and want to reassure you that your data is handled in a safe and secure manner.

All Acquia DAM AI features are opt-in and disabled by default. These features use pre-trained models. Customer data such as image assets and metadata are not used to train AI models.

See the Acquia DAM Product & Services Guide for more details. You can find general AI guidelines for all Acquia products at the Acquia AI User Guidelines page.

AI-Powered Features in Acquia DAM

Acquia DAM offers several AI-powered features designed to enhance productivity and streamline asset management. These features are distinguished with the icon on the Feature Administration page in the Admin app. Each of these features is optional and must be explicitly enabled by an administrator.

1. Facial Recognition

This feature automatically detects and groups faces in your image assets, enabling administrators to efficiently manage people tagging, improve content findability, and maintain compliance with data privacy laws.

• AI Technology Used: The facial recognition feature is powered by Amazon Rekognition.
• How it Works:
  • When enabled, the system detects faces in newly uploaded images. For existing assets, an administrator can manually trigger the process.
  • Amazon Rekognition detects up to 100 faces per image and creates a mathematical representation of each face (a "face vector"), not an actual image of the face.
  • These face vectors are stored in a secure, customer-specific "collection". Detected faces are then compared against existing faces in your collection to group similar faces together.
  • Administrators can then assign names to the grouped faces, which makes those individuals searchable within the DAM.

2. Computer Vision Keywords (Auto-Tagging)

This feature automatically generates descriptive keywords based on the visual content of an image asset, offering a more automated way to categorize assets and make them more discoverable.

• AI Technology Used: This functionality is powered by Amazon Rekognition's DetectLabels feature.
• How it Works:
  • The DetectLabels service analyzes an image to detect real-world entities, events, and concepts like "tree," "wedding," or "landscape".
  • It can also detect image properties like dominant colors and image quality (brightness, sharpness, contrast).
  • The service returns a list of labels with confidence scores, which are then added to the asset as "AI Tags" to improve searchability.

3. AI-Generated Video Transcription

This feature provides a more automated and faster way to generate transcripts of speech in video files, which can then be viewed, downloaded, and searched within Acquia DAM.

• AI Technology Used: This feature uses OpenAI's Whisper, an advanced automatic speech recognition (ASR) solution. Whisper is run within the Acquia private AWS environment; there are no external calls made to OpenAI.
• How it Works:
  • When initiated by a user, the video file is processed by the Whisper model to generate a text transcript of the spoken words.
  • The process is asynchronous, meaning it runs in the background, and the time to completion depends on the length of the video.
  • Once generated, the transcript is stored securely and indexed, making the content of the video searchable.

4. Acquia Copilot

Acquia Copilot is an intelligent assistant that helps users with documentation and support requests, and the ability to summarize and analyze document assets through a natural language interface.

• AI Technology Used: The chatbot is built on Amazon Bedrock, a service that provides access to high-performing foundation models. The model family used is Claude Sonnet from Anthropic, which was chosen for its safety, contextual awareness, and performance. Model versions are updated regularly, so a specific model number is not listed here.
• How it Works:
  • The chatbot processes natural language queries to understand user intent.
  • It then interacts with a support documentation knowledge base contained with the private Acquia AWS environment and DAM's public APIs to fetch relevant data and generate a response.

Security, Data Privacy, and Data Flow

The security, durability, and availability of your data are our top priorities. The entire Acquia DAM platform is built on Amazon Web Services (AWS), leveraging its robust security and compliance capabilities.

Core Security Principles

• Secure by Default: AI features are opt-in only. Your assets will not be processed by any AI service unless you explicitly enable the feature.
• No Training on Your Data: We use pre-trained AI models from providers like AWS and OpenAI. Your data, including assets and metadata, is never used to train or retrain these models.
• Data Isolation: Your data is processed in a secure, isolated environment. For features like facial recognition, each customer has a separate, dedicated "collection" within Amazon Rekognition to ensure faces are never shared or compared across different customer accounts.
• Encryption: All customer data stored in our primary storage, Amazon S3, uses Server-Side Encryption with Amazon S3-Managed Encryption Keys (SSE-S3). All network connections, from the load balancer to the application and database servers, are also encrypted.
• Access Control: We use AWS Identity and Access Management (IAM) to enforce strict, role-based access controls, ensuring that only authorized personnel have access to the underlying infrastructure for maintenance purposes.

Data Flow Diagram

The following diagram illustrates the high-level data flow when an asset is processed by an AI service.

Document ChangeLog