AI Tools
This collection invites exploration: the tools are each specialized for specific tasks within the field of generative AI. All of them operate in compliance with data protection regulations when using the HU's Large Language Models (LLMs). In addition to the pre-configured AI tools, the HU's LLMs can also be integrated into self-developed AI applications via an API. An OpenAI-compatible interface is provided for this purpose. All data remains within the HU's domain.
AI Tools
Overview
|
Interactive chat interface for general AI tasks |
||
|
Text editing without prompting via menu selection |
||
|
Translation of documents and long texts |
||
|
Upload multiple files and interactively query them |
||
|
Speech-to-text transcription with email delivery |
||
|
Structure spoken language into organized text formats |
||
|
Image analysis and visual information extraction |
Specialized Tools
|
Generator for Data management plans |
||
|
Assistant for Forschungsdatenmanagement (Research Data Management) |
Chat
A clear and intuitive chat application suitable for a wide range of tasks
Basic actions are pre-formulated (e.g., translate, improve, summarize). The HU system prompt and response style can be customized. Results can be easily copied. Upon closing the browser window, all inputs and responses are completely deleted.
https://ki-tools.hu-berlin.de/chat/
Text editing with tools
After pasting a text, twelve basic actions can be triggered with a single click. The results are displayed in a side-by-side comparison for easy evaluation. No prompting is required, but additional custom commands can be added. Results can be conveniently exported.
https://ki-tools.hu-berlin.de/text/
Translations
This tool enables translations even of lengthy texts. Specific term translations can be predefined via the Glossary field.
https://ki-tools.hu-berlin.de/translate/
Working with multiple documents
Upload a group of documents you wish to analyze collectively. After technical processing, you can pose questions or assign tasks via prompts. Enable source citations to see which sections of the documents the AI refers to in its responses.
https://ki-tools.hu-berlin.de/doc
Transcription of video and audio files
This tool allows you to transcribe spoken language from audio and video files and output it as text. The extracted transcriptions are automatically sent to your HU email address after processing (as plain text and with timecodes). Uploaded media files can be up to 4 GB in size. All files are automatically deleted in compliance with data protection regulations after processing.
Based on the KI model "Whisper", two modes are offered The slow mode also provides translations; to use this, select the target language for the translation. The fast mode should only be used if the transcription is subsequently further processed by an LLM (see below); otherwise, the quality is insufficient for a pure transcription.
https://ki-tools.hu-berlin.de/stt
Post-processing of Transcriptions
To convert spoken language into conventionally readable text, we offer an additional tool. In the first stage, it linguistically refines the direct transcription while remaining close to the original. In a second stage, the text can be thematically structured and condensed. In a third stage, a formatted document (in Markdown) with automatically generated subheadings is created, which can be immediately reused—for example, as a meeting minutes report.
https://ki-tools.hu-berlin.de/stt-helper
Extracting Information from Images – Machine Vision
Using a Vision Language Model (VLM), images can be analyzed and information extracted from them. To do so, provide a link to the image (shareable links from HU-Box are also acceptable) or upload it directly. Then, specify your task or question via a prompt.
The available prompt gallery provides examples from the fields of image analysis and text recognition, illustrating the various ways in which information can be analyzed or processed.
https://ki-tools.hu-berlin.de/vision
Creating Data Management Plans
Based on curated, discipline-specific examples of real Data Management Plans (DMPs), draft DMPs can be generated. Researchers enter their project-specific details via a structured form, and the system transforms this information into a fully elaborated, clearly structured, and content-coherent DMP draft.
https://ki-tools.hu-berlin.de/dmp
Research Data Management Assistant
The FDM Assistant provides support in research data management. It is based on a curated knowledge base that consolidates institutional guidelines, discipline-specific recommendations, best practice examples, and internal information resources to generate context-specific responses. The system covers all aspects—from project planning and storage and documentation to legal and organizational issues.