Artificial Intelligence was the keyword of last year, we tried it all times and were impressed - but what now? We see AI and Generative AI not as a keyword that we like to write to the flags, but as a useful tool for our work and your websites.
Application examples for websites (Drupal)
- Proposals and suggestions for improvements for texts and text sections
- Creation of texts in Easy Read
- Chat bots
- LLM assistance systems instead of a classic search
- Generating image descriptions and alternative texts for accessibility
- Generating tags and taxonomy trees
- Translation assistance or translation
Independent AI servers
We offer digital sovereign AI servers on an open source basis with user interface, rights management, multiple LLMs, own AI Agents (e.g. for reading websites or external databases) and vector databases as turnkey product.
The AI server provides chat bots that you can insert as snippet into your existing website or your intranet.
With the AI server, you can independently store data and files and make available for your own search or chat bots.
You can control all prompts, data streams and queries and thus make usable for your own organization, for example as an internal knowledge base, search tool or to create texts tailored to your organization.
Digital sovereignty
In the sense of digital sovereignty, we operate our independent AI servers and use different LLMs for the respective application purpose.
We operate servers for vector databases so that data for the retrieval of AI systems can be stored with us data sovereignty, for example for your own RAG application.
We operate self-hosted workflow tools to address even complex AI processes including agents and MCP servers and clients.
Thanks to the APIs of the Drupal modules, external proprietary systems can also be integrated with little effort at the customer's request.
We take care to use AI systems responsibly and in accordance with the AI Regulation (EU) and are also delighted to discuss the ethical aspects with you.
AI Glossary
-
Artificial intelligence (AI) refers to the ability of IT systems to perform tasks that normally require human thinking. These include solving complex problems, identifying patterns and making decisions based on data. AI systems use algorithms and machine learning to learn from large amounts of data and make predictions.
-
Systems with Generative Artificial Intelligence (Generative AI, GenAI) are able to create new content that resembles human products. These can be texts, translations, images, music and other media. Special models are required for the respective application, which have been trained for this purpose. There are also special models that can be used to generate program code.
-
A Large Language Model (LLM) is a language model that is trained with large amounts of text data to understand and generate natural language. LLMs use machine learning and neural networks to identify patterns in language and to create meaningful answers or texts based on them. They are used in various applications, including chatbots, translation services, text processing and artificial intelligence assistants.
-
Retrieval Augmented Generation (RAG) is an approach in artificial intelligence, in which a language model is supplemented by additional information from an external database or a knowledge memory. Unlike traditional language models based only on their internal knowledge, RAG uses a retrieval component to retrieve relevant information and then involve it in the generation process. This allows the model to provide more precise and more up-to-date answers, especially for questions requiring specific knowledge.
Using RAG, for example, a chat bot of a website with additional information can be fed from the databases of the website.
-
Semantic search is an approach to searching for information that aims to understand the intention and context behind a search request instead of searching for keywords. In the context of artificial intelligence (AI), semantic search uses advanced algorithms and machine learning to analyze the meaning of words and sentences in a request. This allows the system to deliver more relevant and more precise results, as it takes into account not only the correspondence of keywords but also the semantic relationship and the intention of the request.
Technical basis of a semantic search is the use of LLMs and vector databases.
-
Vector databases are specialized database systems to efficiently store and query data as high-dimensional vectors. Unlike traditional databases, vector databases are optimized to carry out similarity queries. They use algorithms such as k-nearest neighbors (k-NN) or Nearest Neighbors (ANN) to find the next neighbors of a given vector. These databases are particularly useful in applications based on machine learning and artificial intelligence, such as image and speech recognition, recommendation systems and natural speech processing.
The conversion of data into vector data (the so-called embedding) occurs with specialized LLMs.
-
AI agents are systems that can independently perform tasks in a specific environment. They use artificial intelligence (AI) and machine learning to make decisions and make actions. For example, AI agents can automatically check texts or post them into social media platforms.
-
Workflow tools can be used to implement complex processes without complex programming. Examples: the processing and transfer of web form inputs to external services or a multi-step response of search requests by LLM and several databases.
-
The Model Context Protocol (MCP) is the new standard with which AI agents and other system components can communicate with each other. This makes it possible to create agents with various tasks.
Modules are available for Drupal to act as MCP server or as MCP client.