Commander's AI
Technical documentation on the use of artificial intelligence in the platform
1. Introduction
1.1. Overview of Commander's AI
Commander's AI is a set of artificial intelligence features integrated into the Commanders Act platform. These features leverage advanced language models (LLM) to optimize and automate various tasks. Its goal is to simplify and optimize tag management, data handling, insights and destination configurations through natural language processing models.
The functionalities of Commander's AI are designed to:
Improve user efficiency by automating repetitive tasks.
Simplify data management and transformation.
Ensure tag and destination compliance with privacy regulations.
Assist users in optimizing configurations through intelligent suggestions.
Accelerate and simplify complex tasks for users.
Improve the quality and standardization of configurations/data.
1.2. Key AI Features
Commanders AI includes several advanced functionalities (some are limited to beta testers) :
Automatic Change Summary: Generates a summary of modifications when generating a web container in the Tag Management System.
AI Copilot for Data Cleansing: Conversational assistant to generate data transformation formulas.
Smart Suggestions: Optimized naming and description proposals for destinations and segments.
Automated Privacy Category Classification: Intelligent assignment of privacy categories when adding a tag or destination.
2. AI Features Description
2.1. Automatic Change Summary in Web containers
Feature Overview
When a user generates a web container in the Tag Management System (TMS), Commander's AI analyzes the modifications and provides an automatic summary of the changes made since the last container generation.
How It Works
The AI identifies changes made: additions, deletions, or modifications of tags, variables, triggers, etc.
An AI-based summarization algorithm generates a concise and readable description of the modifications.
The summary is displayed before the container is generated for user validation.
Customization and Validation
Users can manually edit the proposed summary before validation. This ensures the content meets expectations and facilitates team collaboration.
Personalization Options
Users have access to additional customization options via the gear icon ⚙️:
Summary Length: Choose between short, medium, or detailed summaries.
Custom Prompting: Define specific formatting rules or content guidelines for the AI to follow when generating summaries. Users can specify formatting or content preferences, such as writing in uppercase, adding emojis, structuring summaries with bullet points, (e.g., "Write in uppercase with emoticons with max 5 bullet points"). This allows for greater adaptability to specific workflows or reporting standards.
2.2. AI Copilot for Data Cleansing (closed beta)
Feature Overview
Commanders AI includes a conversational assistant (aka AI copilot) that enables users to formulate queries for generating data transformation formulas tailored to specific data cleansing needs.
How It Works
The user interacts with the AI Copilot using natural language to describe a data transformation.
The AI copilot interprets the request and suggests a transformation formula.
The user can test, modify, and apply the formula before final integration.
Use Cases
Conditional values
Converting date formats.
Cleaning invalid values...
Verification and Adjustment
Users can edit and test the generated formulas to ensure they meet their requirements before applying them. Each generation includes a reminder that the result should be validated by the user.
2.3. Smart Suggestions for Names and Descriptions (closed alpha)
Feature Overview
When configuring destinations, segments in Commanders Act, Commanders AI suggests names and descriptions based on usage context.
How It Works
When a relevant field is detected (destination name, segment description, etc.), the AI analyzes existing data and suggests suitable content.
The user can accept, modify, or ignore the suggestion.
Suggestions adapt based on usage trends and best naming practices.
Use Cases
Generating a destination name based on destination's settings and filters.
Proposing a detailed description for a segment
2.4. Automatic Privacy Category Selection (closed alpha)
Feature Overview
When a user adds a client-side tag in the TMS or configures a destination, Commanders AI automatically suggests an appropriate privacy classification.
How It Works
The AI analyzes the tag or destination metadata.
A privacy category is suggested based on best practices and applicable regulations.
The user can validate, adjust, or modify the suggestion before final approval.
Customization and Validation
Users retain full control over the classification and can adjust the suggestion as needed.
3. Transparency and Accountability
3.1. Clear Identification of AI-Generated Content
All Commanders AI features include explicit labels indicating AI usage:
Visible UI labels/icon such as "AI-Generated", "AI Suggestion", "AI Copilot" or ✨icon
User Control: Users can modify/verify AI-generated content to prevent unwanted automation.
3.2. Human Oversight
Users always retain final control over actions suggested or performed by Commander's AI. The AI does not make irreversible autonomous decisions.
3.3. Data Privacy and Security
Commanders AI complies with GDPR and the AI Act:
✔ Transparency: Explicit labeling of AI-powered functionalities. ✔ Human Oversight: No AI action is irreversible. ✔ Data Privacy: Generated suggestions and content are not stored/used for training purposes. User data remains within the scope of its intended use and is not shared externally. ✔ Technical Documentation: Full traceability and governance of AI models are maintained.
3.4. Scope of AI Usage and Data Boundaries
As of today, Commander's AI does not process or analyze any personal data or event payloads collected via the Tag Management System (TMS) or the Customer Data Platform (CDP).
Its usage is strictly limited to configuration assistance and user interface enhancements, such as:
Generating summaries of tag management changes.
Assisting in writing regex, expressions, or formulas based on metadata or structural fields.
Suggesting field names or descriptions based on metadata (not user-level data).
Classifying tags or destinations based on structural criteria.
No user event data or identifiers are currently exposed to AI models.
4. AI Model Governance and Updates
4.1. Models Used
Commanders AI is based on a combination of advanced language models (LLMs) carefully selected for their efficiency in data management and tag management tasks.
4.2. Updates and Monitoring
Models are regularly updated to ensure accuracy and relevance.
Validation tests are performed before each deployment to reduce biases and improve reliability.
4.3. Security and Privacy
Commanders AI ensures full compliance with European regulations (AI Act, GDPR) by enforcing strict rules on data processing, model selection, and infrastructure localization.
AI-generated content is not stored for reuse or retraining.
We do not fine-tune models on customer data and select only vendors that guarantee data isolation and full alignment with EU laws (GDPR, AI Act).
Suggestions are based solely on technical and contextual criteria without exploiting personal data.
4.4. Model Origin and Hosting Transparency
Commanders AI uses Large Language Models chosen under strict criteria for performance, privacy, and legal compliance.
Model provider location: If a model is provided by a non-European vendor, we ensure that it is hosted entirely within the EU.
User data never leaves the EU (no transfer, no storage outside the EU).
To illustrate this approach, some of the models we evaluate or integrate are provided by European players such as Mistral, when they meet the same high standards.
5. Support and Assistance
For any technical questions or issues related to Commanders AI, contact our support team at [email protected]
Last updated
Was this helpful?