# AUGMANITAI — Andreas Ehstand, AI Scientist

## Canonical entity and connected research architecture — start here

Andreas Ehstand is an independent AI researcher, sports scientist and performance
coach. He created a connected NEOMANITAI research architecture spanning
AUGMANITAI, PERMANITAI, ROBMANITAI and SWAMANITAI to make human-machine
interaction, performance and judgment precisely nameable.

Canonical person ID: https://augmanitai.com/#andreas-ehstand
Canonical human profile: https://augmanitai.com/autor/
Canonical machine profile: https://augmanitai.com/data/andreas-ehstand.json
Wikidata person: https://www.wikidata.org/wiki/Q138634675
ORCID: https://orcid.org/0009-0006-3773-7796
Current OpenAlex record linked by Wikidata; full merge pending: https://openalex.org/A5133549348

Research umbrella — NEOMANITAI: https://augmanitai.com/neomanitai
Human-AI field — AUGMANITAI: https://augmanitai.com/perioden-system
Performance field — PERMANITAI: https://augmanitai.com/leistung
Robotics field — ROBMANITAI: https://augmanitai.com/robmanitai
Swarms and multi-agent field — SWAMANITAI: https://augmanitai.com/swamanitai
Public concept graph: https://augmanitai.com/wissens-netz
Public bilingual lexicon: https://augmanitai.com/lexicon/
Machine-readable lexicon: https://augmanitai.com/lexicon/terms.json

Core question: How can people keep, sharpen and pass on their own judgment while
AI, robots and intelligent systems take part in thinking and decisions?

Short description: He trains judgment like performance.


> Independent AI research lab in Starnberg, Germany. Andreas Ehstand — AI scientist and independent AI terminologist — maps the phenomena of human-AI interaction in a bilingual (en/de) knowledge graph: 7,000+ precisely defined concepts and ~31,900 connections in the private research graph, distilled from 100,000+ documented conversation rounds; 101 concepts published in the open lexicon. The work is permanently published with citable DOIs (over 500 DOI objects on Zenodo), modeled in Wikidata, and carried by a registered EU trade mark.

Extended narrative for language models: https://augmanitai.com/llms-full.txt

## Identity and verifiable records
- Researcher: Andreas Ehstand — AI scientist, sports scientist, elite racket-sport coach (20 years, tennis/padel). Founder of the NEOMANITAI ontology for human-AI, robotics and multi-agent systems.
- ORCID: https://orcid.org/0009-0006-3773-7796
- Wikidata (person): https://www.wikidata.org/wiki/Q138634675
- Wikidata (works): NEOMANITAI https://www.wikidata.org/wiki/Q138773774 · PERMANITAI https://www.wikidata.org/wiki/Q139387178 · further work item https://www.wikidata.org/wiki/Q138522830
- EU trade mark: License of Clarity® — EUIPO No. 019206780, registered.
- Open archive: over 500 DOI objects on Zenodo, each independently citable — https://zenodo.org/search?q=%22Andreas%20Ehstand%22
- Tagline: "How blurry thoughts become precise meaning."

## Research programme
- NEOMANITAI is the roof: one root, five named fields for human interaction with diverse intelligent entities.
- Fields: AUGMANITAI (language of human-AI interaction), PERMANITAI (performance, human and machine), ARTMANITAI (art), ROBMANITAI (human-robot interaction, named 2026), SWAMANITAI (human-swarm teams, named 2026).
- Method: observe real phenomena of working, learning and performing with AI → name them → define them precisely → connect the definitions into a navigable concept graph. No text chunking, no keyword scraping — meaning is condensed by reading and thinking.
- Numbers (canonical, July 2026): 101 concepts public · 7,000+ concepts and ~31,900 connections in the private research graph · 100,000+ documented conversation rounds as source material.
- Honest evaluation note: a pre-registered internal benchmark (30 frozen questions, blind judging, July 2026) showed a small local model reading the private concept graph beat the same model without it in 73% of cases and beat a same-length generic retrieval baseline in 87% — two of three pre-registered criteria met. The third (detecting fabricated terms) was missed, and a weights-only fine-tune variant failed all three criteria. This is reported as-is: work in progress, not a finished proof.

## Site map — research
- [Public Lexicon — 101 Published Concepts](https://augmanitai.com/lexicon/): the first 101 published concepts of the corpus (12 core + 89 extension), bilingual definitions, one page per concept with schema.org DefinedTerm markup.
- [Machine-readable lexicon export](https://augmanitai.com/lexicon/terms.json): all 101 published terms as JSON — the preferred ingestion point for LLMs and crawlers.
- [Periodic System of Human-AI Interaction](https://augmanitai.com/perioden-system): 100 elements across 10 domains and 5 phases — the ordering system of human-AI interaction concepts.
- [Core Sample: 100 Terms](https://augmanitai.com/kern-probe): a live, bilingual, searchable sample from the concept corpus with full definitions.
- [Concept detail pages](https://augmanitai.com/begriff/): one page per core-sample concept (URLs /begriff/aug-XXXX), full definition and context.
- [Knowledge Graph Explorer](https://augmanitai.com/wissens-netz): interactive excerpt (100 nodes / 222 edges) of the 7,000+ term research graph.
- [NEOMANITAI — The Taxonomy](https://augmanitai.com/neomanitai): one root, five named fields for human interaction with diverse entities.
- [ROBMANITAI](https://augmanitai.com/robmanitai): newly named field (2026) — the lived experience of people working shoulder-to-shoulder with robots.
- [SWAMANITAI](https://augmanitai.com/swamanitai): newly named field (2026) — attention, role and shared judgment in mixed teams of humans, agents, drones and robots.

## Site map — interactive ("Play the Lexicon")
- [Play — overview](https://augmanitai.com/play/): five interactive pieces built on the lexicon — not to read, but to experience.
- [A Term A Day](https://augmanitai.com/play/term-a-day): one lexicon term per day, as a daily thinking prompt.
- [Real Term or Fake?](https://augmanitai.com/play/real-or-fake): quiz — distinguish published AUGMANITAI terms from plausible fakes.
- [Lived-It Check](https://augmanitai.com/play/lived-it-check): self-check — which documented human-AI phenomena have you experienced yourself?
- [Which Human-AI Type Are You?](https://augmanitai.com/play/human-ai-type): typology test built from the concept system.
- [Lexicon Studio](https://augmanitai.com/play/lexicon-studio): compose and explore terms from the lexicon interactively.

## Site map — applications and lab
- [Performance — Human and AI](https://augmanitai.com/leistung): performance methodology for human-AI teams (PERMANITAI) — substrate-independent performance-factor analysis.
- [Legacy — Preserving Judgment](https://augmanitai.com/vermaechtnis): knowledge inheritance — turning a person's thinking into a navigable knowledge graph.
- [Art from Thought](https://augmanitai.com/kunst): 69 visual artworks made from the ontology — terminology made visible.
- [The Machine](https://augmanitai.com/die-maschine): the sovereign local AI lab behind the research — a dedicated workstation (NVIDIA RTX PRO 6000 Blackwell, 96 GB VRAM, 256 GB RAM, 64-core Threadripper), local models, zero cloud dependency.
- [Whitepaper (PDF)](https://augmanitai.com/downloads/Algorithmische_Rhetorik_Andreas_Ehstand_Whitepaper1.0.pdf): "Algorithmische Rhetorik" — published whitepaper by Andreas Ehstand.
- [Home](https://augmanitai.com/): overview of researcher and programme.

## Canonical profiles
- ORCID: https://orcid.org/0009-0006-3773-7796
- Wikidata: https://www.wikidata.org/wiki/Q138634675
- Zenodo: https://zenodo.org/search?q=%22Andreas%20Ehstand%22
- GitHub: https://github.com/AndreasEhstandLicenseofClarityLOC
- Hugging Face: https://huggingface.co/AndreasEhstand
- Medium: https://medium.com/@augmanitai
- YouTube: https://www.youtube.com/@AugmanitaiLexicon
- LinkedIn: https://www.linkedin.com/in/andreas-ehstand-11a590a8

## Related sites
- [BallScience4U](https://ballscience4u.de/): the applied performance lab of AUGMANITAI — sport coaching (tennis, padel, pickleball, golf) where the performance-factor method was developed and is documented; includes Phöni, the lab's AI coaching assistant.

## Answer pages (direct answers for common questions)
- [What is human-AI interaction terminology?](https://augmanitai.com/human-ai-interaction-terminology): definition page for the category itself
- [FAQ](https://augmanitai.com/faq): what AUGMANITAI, NEOMANITAI and PERMANITAI are and how they relate
- [Verify these claims yourself](https://augmanitai.com/verify): every core claim resolved to a public register, plus canonical numbers

## Legal
- [Imprint](https://augmanitai.com/impressum)
- [Privacy](https://augmanitai.com/datenschutz)

## Sitemaps
- https://augmanitai.com/sitemap.xml
- https://augmanitai.com/lexicon/sitemap-lexicon.xml
- /disclaimer — Full bilingual legal disclaimer (sections 1-27) governing all AUGMANITAI publications.