LLM-Enabled Development
This site documents a growing collection of AI-powered tools built on locally operated Large Language Models. All tools were developed as rapid prototypes and are continuously being improved.
Explore the Tools
Archive & Insights
Writing & Text
| Tool |
Description |
| Chat |
Interactive AI chat with model profile system, document analysis and image processing |
| Text Workshop |
Agentic text processing with task profiles, hallucination checks and versioning |
| TextTool |
Iterative text editing with predefined functions (legacy version) |
| Style Tool |
Transforms texts according to predefined style profiles with structured profile schema |
| Text Style Editor |
Slider-based style editor with 34 parameters, neutralization and presets |
| Translate |
Translates documents while preserving structure, formatting and glossary |
Speech & Audio
| Tool |
Description |
| TTS Suite |
Multi-voice audio production: podcasts, interviews, audiobooks with three TTS model families and ASR quality assurance |
| ASR Transcription |
Speech recognition with audio preprocessing, segmentation, hallucination cleanup and subtitle export |
| Speech-to-Text |
Three-stage transcript processing via LLM pipeline (legacy version) |
Images & Visualization
| Tool |
Description |
| Image Generator |
Text-to-image and image-to-image with preset system, iterative editing and two image models |
| Diagram Generator |
Agentic pipeline for Mermaid and draw.io diagrams with validation and correction loop |
| Grafix |
Chat-based business diagram editor with deterministic template engine and 27 diagram types |
| Chart Generator |
Agentic data visualization from CSV/Excel with intent detection, planning phase and fallback logic |
Research & Analysis
| Tool |
Description |
| Research Tool |
Agentic multi-stage research with dual LLM, gap analysis, contradiction detection and academic databases |
| TalkToDocuments |
Long-context document analysis with up to 250,000 tokens, source references and preprocessing pipeline |
| Code Analyzer |
Multi-stage code analysis with four LLM roles, YAML intermediate layer and aggregated system evaluation |
| Vision |
Image and PDF analysis with Vision Language Models, page selection and alt text generation |
Documents & Presentation
| Tool |
Description |
| PPT Workshop |
Agentic slide deck creation with five-phase pipeline, layout catalog and quality assurance |
| Presentation Prep |
Two-agent architecture for slide structuring from documents (legacy version) |
| CodeDocumentation |
Generates complete documentation sets from repositories with three-stage pipeline and framework extractors |
| Doc |
Automatic document summarization (legacy version) |
| Web Helper |
Migrates and transforms web content between CMS systems |
| Plone Migration |
Processing and LLM-assisted revision of Plone CMS exports with hierarchy reconstruction |
Data Collection
| Tool |
Description |
| Flex Mapping |
Configurable information extraction system with QA layer, confidence scores and static website export |
| AI Mapping |
Web-based mapping and data extraction |
| AI Survey |
Multi-stage LLM pipeline for open surveys with clarity assessment, follow-ups and structuring |
| Survey Analysis |
Structured survey data analysis with LLM-powered clustering, segmentation and report generation |
| Staff Cost Calculator |
TV-L personnel cost calculation from natural language with deterministic computation and Excel/PDF export |
Infrastructure & Operations
| Tool |
Description |
| LLM Manager |
Management UI for vLLM containers on GPU servers: 7 model types, API routing, MIG, monitoring and load balancing |
Archive
The original experiment documentation including comics, detailed
workshop reports and cross-cutting insights can be found in the Archive.