CI in Content Management Systems

1. Native KI-Features in the CMS Core

This strategy involves integrating generative AI functions directly into the user interface of the CMS, usually by calling external Large Language Models (LLMs) via predefined REST APIs, such as OpenAI. The models are commonly connected using public API keys and are available in text fields or editor interfaces as buttons or sidebars.

Advantage: Editors can generate, revise, or structure content without training.

Disadvantage: No client capability, no authentication via corporate accounts (e.g. via SSO), no control over whether user inputs are used for model training.

2. Integration of external AI systems via API

The strategy here relies on providing open APIs to integrate proprietary or preferred AI models - for example, hosted instances of Azure OpenAI, Claude, or Mistral. These API endpoints are often configurable, allowing for the inclusion of authentications, logging, or prompt templates. This is frequently implemented in the context of "Bring Your Own Key" (BYOK), enabling companies to store their own API keys and thus maintain full control over usage and billing.

Advantage: Companies retain control over model selection, data protection compliance, and infrastructure.

Disadvantage: Setup and maintenance require technical understanding, API design, and possibly DevOps capabilities.

3. Fully configurable enterprise AI

This strategy relies on deep system integration: AI functions are fully integrated into the CMS itself – including connection to existing company structures such as Azure Active Directory (SSO), company-specific access rights, dedicated infrastructure, and private LLM instances. These are fully controlled workflows in which content, prompts, and outputs remain entirely within the organization.

Advantage: Maximum governance, compliance requirements adherence, and complete separation from other clients or public APIs.

Disadvantage: Significantly higher technical effort and licensing costs; usually only economically feasible for large companies with corresponding infrastructure and IT teams.

TYPO3 is a free and open-source content management system written in PHP.

There is no AI integration in TYPO3 Core yet. However, there are several extensions that deal with this topic—the best known are TYPO3 AI Suite and T3AI. TYPO3 AI Suite aims to make generative AI functions such as text suggestions, meta tag generation, and image captions available directly in the backend. In the standard packages, the extension uses standardized APIs (e.g., ChatGpt and Midjourney) and allows individual configurations, e.g., with regard to prompt templates or language settings. Those who want to use their own API keys can do so as part of the “Bring Your Own Keys” (BYOK) package. This allows company-owned enterprise accounts (e.g., for ChatGPT) to be stored. This provides expanded options, particularly with regard to data protection, data sovereignty, and governance. User input is not automatically transmitted to third parties for training purposes – the behavior depends on the connected model and the API conditions of the provider used. Operators can thus configure AI services in a data protection-compliant and isolated manner.

Sitecore

Sitecore's Composable DXP is a collection of various components (CMS, personalization, search, customer data platform, marketing automation, etc.) that can be combined to perform virtually all digital marketing tasks from a single platform. AI is used to support the respective tasks in virtually all of these modules. In addition, Sitecore Stream provides a dedicated AI layer in Sitecore that all components can access. Sitecore Stream can be equipped with your own documents about your portfolio, brand, target groups, product details, and much more, and then provides brand-aware AI, generative copilots, and agentic workflows that are embedded in products such as Content Hub or XM Cloud. They support content creation, campaign orchestration, and A/B testing directly in everyday marketing. Customers can store their own enterprise accounts (e.g., Azure OpenAI) – including SSO integration and individual authentication. Sitecore emphasizes that content and prompts remain in the enterprise context and are not used for training public models. The architecture is clearly focused on data protection, governance, and regulatory compliance. Sitecore's vision: to shift all marketer tasks to the Sitecore world—from the creation of creative briefings to content production to personalized delivery across various channels.

Ibexa

Starting from version 4.6.12, Ibexa AI Actions can be installed. The package includes a framework and the OpenAI Connector. After setup, editors can optimize selected content (e.g. text sections or image descriptions) directly in the editor or generate ALT texts. Extensions through additional AI services or custom handlers are possible, as well as the integration of local AI installations, such as LLaMA. Ibexa Connect allows AI Actions to be embedded in automation scenarios, integrated into workflows, and triggered via the UI - without the need for programming.

With Ibexa DXP V5 (Release September 2025), the integration of AI Actions is set to be significantly enhanced: According to the announced innovations, scenarios from Ibexa Connect should be directly accessible in the CMS in the future. The support of, among others, Claude, Le Chat, or internally hosted models is planned.

Storyblok

Storyblok pursues an editorial-driven strategy: the AI features—including text generation, text optimization, and translations—are native and available directly in the visual editor without any additional setup. This allows editors to edit content in context or create translations without leaving the editing environment. The “Custom AI Features” function and individual API keys also allow users to connect their own models or company accounts (e.g., Azure OpenAI). This makes it possible to implement specific data protection and governance requirements. In addition, the Storyblok App Hub offers various AI plugins (e.g., for SEO optimization, automatic meta data generation, image generation, or automatic image description) that can be flexibly integrated into existing workflows. In addition, Storyblok offers the Ideation Room, a collaborative environment where teams can jointly develop content ideas, provide feedback, and incorporate AI-supported suggestions in real time. This supports and accelerates the creative process directly in the CMS. Overall, the focus of AI use at Storyblok is clearly on fast, convenient use for editorial teams, combined with optional extensions for companies with higher compliance requirements.

Kentico

Kentico integrates AI through the Aira module, which is available in the Advanced licenses and specifically supports marketing and content teams. Aira offers features for content optimization for images (e.g. setting focus points, generating automatic ALT text) and texts (text revision, automatic text generation for emails). In addition, Aira analyzes user behavior and interaction data to identify trends or potential issues in the customer journey. Based on these analyses, Aira can automatically segment target audiences and deliver personalized content or offer recommendations.

The recommendations from Aira are based on machine learning models that identify relevant content for each user group. The solution uses Azure OpenAI as its technological foundation and is deeply integrated into the Kentico Digital Experience Platform, allowing Aira to also be connected to CRM or e-commerce systems. Currently, it is not possible to store your own enterprise accounts for AI providers, which is relevant for organizations with high compliance requirements. However, according to the roadmap, Kentico plans to expand Aira with additional features in the future, particularly to support translation processes.