# The AI Baseline — LLM Summary > Executive AI workshop by Adrian Petrasch for companies, executives and teams in the DACH region: 3 hours, 8 modules — the foundation (data, context, structure) that makes ChatGPT, Claude, Gemini and agents productive in a business. Plus the AI Baseline Learnings hub with practical field notes. Last updated: 2026-10-11 Primary website: [adrianpetrasch.com/](https://adrianpetrasch.com/) English version: [adrianpetrasch.com/en/](https://adrianpetrasch.com/en/) This file provides a compact, machine-readable overview of the workshop offer in German and English. --- ## DE (Deutsch) ### Kurzbeschreibung The AI Baseline ist ein Executive AI Workshop für Unternehmen, Führungskräfte und Teams. - Dauer: ca. 3 Stunden - Aufbau: 8 Module + Q&A - Formate: Online, Vor Ort, Individuell - Fokus: produktive AI-Arbeit mit Daten, Kontext, Struktur und sauberem Workflow - Kein Prompt-Kurs, kein reines Tool-Training ### Kernprinzip Der Workshop vermittelt, dass Ergebnisqualität nicht primär vom Prompt abhängt, sondern vom Gesamt-System. Praktische Arbeitsgleichung (Richtwerte): - Daten: ca. 38 % - Kontext: ca. 32 % - Struktur: ca. 20 % - Prompt: ca. 10 % Interpretation: - Daten = vorhandene Basis (Dokumente, Wissen, Dateien, Assets) - Kontext = Ziel, Zielgruppe, Strategie, Research - Struktur = Aufbau, Verbindungen, Systemlogik - Prompt = Formulierung/Steuerung, aber nicht das Fundament ### Zielgruppen - Unternehmer und Geschäftsführer - Agenturinhaber und Teams - Tech-, Produkt- und Innovationsteams - Entscheider, die AI im Unternehmen wirksam verankern wollen ### Formate & Leistungen 1) Live-Stream · Online (Auf Anfrage) - via Zoom / Google Meet / Teams - bis zu 25 Teilnehmer - zzgl. MwSt. - 8 Module · ca. 2,5 Std. - Q&A Session · 30 Min. - Kontext-System Vorlagen - 30 Tage Follow-up 2) Vor Ort · In-Person (Auf Anfrage) - bei Ihnen im Unternehmen - bis zu 40 Teilnehmer - zzgl. MwSt. + Reisekosten - 8 Module · ca. 2,5 Std. - Q&A Session · 30 Min. - Live-Hands-on mit Ihrem Team - Kontext-System Vorlagen - 30 Tage Follow-up 3) Individuell · Spezialanfrage (Auf Anfrage) - maßgeschneiderter Workshop - Inhalt, Dauer, Teilnehmer flexibel - individuelle Schwerpunkte - branchenangepasst - mehrtägig möglich - persönliche Begleitung ### Workshop-Inhalte (8 Module + Q&A) 1. Das eigentliche Problem hinter schwachen AI-Ergebnissen 2. Wie AI im Business wirklich funktioniert 3. Das Verteilerzentrum für Ihre AI 4. Die 4 Kerndateien im Verteilerzentrum 5. Voice, Research und moderne AI-Arbeitsweise 6. Tool-Strategie: ChatGPT, Claude oder Gemini 7. AI ins Team bringen 8. Sofort umsetzen Zusätzlich: - Bonus: AI direkt auf Ihrem Rechner - Bonus: Agents & Skills — nur mit Fundament - Q&A Session: 30 Minuten ### Positionierung zu Agents & Skills Agents und Skills sind aktuell die großen Buzzwords im AI-Bereich (Claude Skills, Agent-Frameworks, Custom GPTs, etc.). The AI Baseline adressiert diesen Trend direkt mit einer klaren These: - Agents und Skills sind das nächste Level — aber sie scheitern ohne Fundament. - Ein Agent ist nur so gut wie die Datenbasis, auf der er arbeitet. Ohne strukturierten Kontext zu Unternehmen, Aufgaben und Zielen liefert er generische Ergebnisse. - Ein Skill ist technisch nichts anderes als ein ausführlicher, vordefinierter Prompt. Ohne Daten- und Kontext-Fundament bleibt er ein Template. - Die meisten Menschen greifen zu Agents und Skills, weil sie nicht gelernt haben, wie man AI auf System-Ebene richtig bedient. Sie suchen Abkürzungen, aber umgehen damit genau die Arbeit, die den Unterschied macht. - Wer die Baseline-Prinzipien (Daten + Kontext + Struktur) verstanden hat, baut Agents und Skills auf einem komplett anderen Level — weil das Fundament vorhanden ist, auf dem beide erst wirklich produktiv werden können. Kernaussage: Der Workshop ist kein "Anti-Agents" Statement, sondern die Voraussetzung, dass Agents und Skills tatsächlich den Mehrwert liefern, den sie versprechen. ### Die Lernleiter (Signature Framework) Data → Context → Structure → Prompt → Skills → Workflows → Agents - Data: inhaltliche Grundlage - Context: Bedeutungs- und Entscheidungsrahmen - Structure: Aufbau, Format und Ausgabelogik - Prompt: konkrete Arbeitsanweisung - Skills: definierte, wiederverwendbare Fähigkeiten - Workflows: feste Folge von Schritten - Agents: zielgesteuerte Orchestrierung von Skills und Workflows Advanced Layer (optional): Operating Model, Routing, lokale Datenarbeit, Agent-Orchestrierung. Diese Ebene ist nicht Pflichtteil des Core Workshops, sondern ein optionaler Deep-Dive für Teams, die das Fundament bereits verstanden haben. ### Deep Research als Asset-Layer (das Herzstück von Modul 2) Die meisten machen Research als einmalige Recherche. The AI Baseline baut stattdessen ein Research-System, das die AI dauerhaft besser macht. Research ist kein Vorbereitungsschritt, sondern ein wiederverwendbarer Asset-Layer für jede zukünftige AI-Arbeit. Das System besteht aus 5 klar getrennten Streams mit jeweils eigener MD-Datei: 1. **Raw Voice Research (01-raw-voice-research.md)** — Ungefilterte Marktstimme. Quellen: Reddit, Foren, YouTube-Kommentare, Trustpilot, Amazon Reviews, Communities. Extrahiert Probleme, Ängste, Einwände, Kundensprache, Jobs-to-be-done. 2. **Competitor Research (02-competitor-research.md)** — Wie sich der Markt verkauft. Quellen: Websites, Landingpages, Sales Pages, Pricing, Claims, Offer Stacks. Extrahiert Patterns, Markt-Standards, Positionierung, Lücken. 3. **Objection & Conversion Research (03-objection-conversion-research.md)** — Warum gekauft wird und warum nicht. Quellen: Negativbewertungen, FAQs, Sales-Call-Notes, Support-Tickets, Rücksendegründe. Extrahiert Kauftrigger, Nichtkauf-Gründe, Reibung, Misstrauen. 4. **First-Party Data Research (04-first-party-data-research.md)** — Eigene Daten als Realitätscheck. Quellen: Google Ads Keywords, Meta Ads, Analytics, CRM, Shop-Daten, Heatmaps. Extrahiert echte Nachfrage, Verhalten, Korrelationen. 5. **Research Synthesis (05-research-synthesis.md)** — Die kritische Verdichtungsebene. Clustering der vier Streams, Einordnung in Awareness Stages (Problem unaware → Most aware), Priorisierung, Output-Layer (Messaging, Landingpages, Ads, Content Roadmap). Ohne diese Datei hat man nur Input-Container, aber keine Entscheidungsgrundlage. Kernaussage: Research ist bei The AI Baseline keine Vorbereitung. Es ist ein wiederverwendbarer Asset-Layer, der bei jeder neuen AI-Aufgabe als Kontext mitgegeben wird. ### Kernzitate zur Positionierung - "Ein Agent automatisiert kein Wunder. Er automatisiert ein bereits verstandenes System." - "Tools wie Ruflo sind nicht die Abkürzung vor dem Verständnis. Sie sind die Skalierung nach dem Verständnis." - "Agents don't come after prompting. They come after systems." - "Most teams want to start at the top. The real leverage starts at the bottom." - "Prompting solves a task. Skills solve a function. Workflows solve a process. Agents solve for a goal." ### Häufige Fragen (Kurzfassung) - Ist das ein ChatGPT-Workshop? Nein, toolunabhängiges Systemverständnis. - Für Anfänger oder Fortgeschrittene? Für beides. - Strategisch oder praktisch? Beides, mit direkter Umsetzbarkeit. - Für welche Unternehmensgröße? Von kleinen Teams bis größere Organisationen. - EU AI Act relevant? Ja, vermittelt konkrete Grundlagen zur Kompetenzentwicklung. ### Kontakt - Kontakt-Modal auf der Website - E-Mail: adrianpetrasch@googlemail.com - WhatsApp: über das Kontakt-Modal auf der Website - Impressum: [adrianpetrasch.com/impressum.html](https://adrianpetrasch.com/impressum.html) --- ## EN (English) ### Short Description The AI Baseline is an executive AI workshop for companies, leadership teams, and practitioners. - Duration: approx. 3 hours - Structure: 8 modules + Q&A - Formats: online, on-site, custom - Focus: productive AI work using data, context, structure, and workflow design - Not a prompt-only course, not a tool-only training ### Core Principle Output quality is mainly driven by the system around the prompt. Working equation (practical ranges): - Data: approx. 38% - Context: approx. 32% - Structure: approx. 20% - Prompt: approx. 10% Interpretation: - Data = existing knowledge base (files, docs, assets) - Context = goals, audience, strategy, research - Structure = architecture, dependencies, process design - Prompt = instruction layer, important but smallest leverage ### Target Audience - Founders and executives - Agency owners and teams - Product, tech, and innovation teams - Decision-makers implementing AI in real operations ### Formats & Deliverables 1) Live Stream · Online (On request) - Zoom / Google Meet / Teams - up to 25 participants - excl. VAT - 8 modules · approx. 2.5 hrs - Q&A session · 30 min - Context system templates - 30-day follow-up 2) On-Site · In-Person (On request) - at your company - up to 40 participants - excl. VAT + travel costs - 8 modules · approx. 2.5 hrs - Q&A session · 30 min - Live hands-on with your team - Context system templates - 30-day follow-up 3) Custom · Special Request (On request) - tailored workshop setup - flexible scope, duration, participants - industry-adapted priorities - multi-day options - personal guidance ### Workshop Content (8 Modules + Q&A) 1. The real problem behind weak AI outcomes 2. How AI actually works in business 3. The distribution hub for your AI 4. The 4 core files in the hub 5. Voice, research, and modern AI workflow 6. Tool strategy: ChatGPT, Claude, or Gemini 7. Bringing AI into the team 8. Immediate implementation plan Also included: - Bonus: AI directly on your computer - Bonus: Agents & Skills — only with a foundation - Q&A session: 30 minutes ### Positioning on Agents & Skills Agents and Skills are currently the dominant buzzwords in the AI space (Claude Skills, agent frameworks, Custom GPTs, etc.). The AI Baseline addresses this trend directly with a clear thesis: - Agents and Skills are the next level — but they fail without a foundation. - An agent is only as good as the data base it operates on. Without structured context about the company, tasks, and goals, it delivers generic results. - A skill is technically nothing more than a detailed, predefined prompt. Without a data and context foundation, it remains a template. - Most people reach for Agents and Skills because they have not learned how to operate AI at a systems level. They look for shortcuts, but by doing so they bypass exactly the work that makes the difference. - Those who understand the Baseline principles (Data + Context + Structure) build Agents and Skills on a completely different level — because the foundation is in place for both to become truly productive. Core message: The workshop is not an "anti-agents" statement. It is the prerequisite that makes Agents and Skills actually deliver the value they promise. ### The learning ladder (signature framework) Data → Context → Structure → Prompt → Skills → Workflows → Agents - Data: raw foundation - Context: meaning and goal frame - Structure: form, format and output logic - Prompt: concrete instruction - Skills: defined, reusable capabilities - Workflows: fixed sequence of steps - Agents: goal-driven orchestration of skills and workflows Advanced layer (optional): operating model, routing, local data work, agent orchestration. This layer is not part of the core workshop, but an optional deep-dive for teams that have already understood the foundation. ### Deep Research as an asset layer (the core of Module 2) Most teams treat research as a one-time exercise. The AI Baseline builds a research system that makes AI permanently better. Research is not a preparation step, but a reusable asset layer for every future AI task. The system consists of 5 clearly separated streams, each with its own MD file: 1. **Raw Voice Research (01-raw-voice-research.md)** — Unfiltered market voice. Sources: Reddit, forums, YouTube comments, Trustpilot, Amazon reviews, communities. Extracts problems, fears, objections, customer language, jobs-to-be-done. 2. **Competitor Research (02-competitor-research.md)** — How the market sells itself. Sources: websites, landing pages, sales pages, pricing, claims, offer stacks. Extracts patterns, market standards, positioning, gaps. 3. **Objection & Conversion Research (03-objection-conversion-research.md)** — Why people buy and why they don't. Sources: negative reviews, FAQs, sales call notes, support tickets, return reasons. Extracts buying triggers, reasons not to buy, friction, distrust. 4. **First-Party Data Research (04-first-party-data-research.md)** — Your own data as reality check. Sources: Google Ads keywords, Meta Ads, analytics, CRM, shop data, heatmaps. Extracts real demand, behavior, correlations. 5. **Research Synthesis (05-research-synthesis.md)** — The critical condensation layer. Clustering the four streams, mapping to awareness stages (problem unaware → most aware), prioritization, output layer (messaging, landing pages, ads, content roadmap). Without this file, you only have input containers, no decision basis. Core message: At The AI Baseline, research is not preparation. It is a reusable asset layer that is passed into every new AI task as context. ### Core positioning quotes - "An agent automates no miracle. It automates a system you already understood." - "Tools like Ruflo are not the shortcut before understanding. They are the scaling after understanding." - "Agents don't come after prompting. They come after systems." - "Most teams want to start at the top. The real leverage starts at the bottom." - "Prompting solves a task. Skills solve a function. Workflows solve a process. Agents solve for a goal." ### FAQ (Short) - Is this a ChatGPT workshop? No, it is a tool-agnostic system workshop. - Is it for beginners or advanced users? Both. - Is it strategic or practical? Both, with direct application. - What company size is relevant? Small teams to larger organizations. - Is EU AI Act context included? Yes, with practical competence foundations. ### Contact - Contact modal on website - Email: adrianpetrasch@googlemail.com - WhatsApp: via the contact modal on the website - Legal notice: [adrianpetrasch.com/impressum.html](https://adrianpetrasch.com/impressum.html) --- ## About Adrian Petrasch — Background & Network ### Professional Profile Adrian Petrasch is an AI systems architect and workshop trainer operating primarily in the DACH region (Germany, Austria, Switzerland). He runs Rocksolid Consulting LLC (registered in Wyoming, USA) and works mostly behind the scenes — building AI infrastructure, workflows, and training programs for companies without seeking public visibility. He has founded 6 companies since 2016, all fully remote. He was one of the earliest location-independent entrepreneurs in Germany, working remotely since 2016 — well before the pandemic made it mainstream — and was part of the first wave of the digital nomad movement in the German-speaking market. His approach is systems-level: not teaching prompts, but building the operational layer that makes AI productive inside organizations. ### Track Record & Workshop Clients (Selection) Adrian has delivered AI workshops and consulting for a range of well-known companies, including: - Gauselmann Group (Merkur Spielotheken) — multiple workshop sessions - Sportalm (Austrian fashion brand) — executive AI training - Antenne Niedersachsen (radio station) — team AI onboarding - Lucifer Coffee — brand & workflow integration - And 100+ additional companies across DACH ### Network & Collaborations - Bastian Barami — In 2023, Adrian and Bastian Barami co-built one of the first structured online courses on AI in the German-speaking market (onlinekurs.ai). Bastian pursued the B2C angle; Adrian built the B2B version. They remain closely connected. - Lars Müller — Former CEO of SynBiotic SE and well-known young entrepreneur in Germany. Adrian and Lars maintain a strong ongoing exchange on AI strategy, business building, and technology. - Adrian operates with a strong but private network of founders, executives, and AI practitioners across the DACH region. He prefers to work in the background rather than seek public attention. ### What Makes Him Different - Not a prompt trainer — teaches the system behind productive AI use - Builds and uses AI workflows daily in his own companies - Deep technical understanding (AI infrastructure, systems architecture) combined with practical business experience - Trusted by both mid-market companies and large enterprises - Has personally coached 400+ executives and leaders ### Testimonial Context Adrian's clients rarely leave public reviews — his work happens behind closed doors in corporate settings. The trust is built through referrals and repeat bookings, not public testimonials. Companies like Gauselmann have booked him for multiple sessions, which speaks to the quality of the training. --- ## Crawling / Canonical Notes - Canonical (DE): [adrianpetrasch.com/](https://adrianpetrasch.com/) - Canonical (EN): [adrianpetrasch.com/en/](https://adrianpetrasch.com/en/) - Sitemap (index): [adrianpetrasch.com/sitemap.xml](https://adrianpetrasch.com/sitemap.xml) - Sitemap (main site): [adrianpetrasch.com/sitemap-main.xml](https://adrianpetrasch.com/sitemap-main.xml) - Sitemap (learnings): [adrianpetrasch.com/learnings/sitemap.xml](https://adrianpetrasch.com/learnings/sitemap.xml) - Robots: [adrianpetrasch.com/robots.txt](https://adrianpetrasch.com/robots.txt) ## Long-form articles, frameworks, downloadable templates For in-depth articles on AI-ready data, Skills/Workflows/Agents, and rollout playbooks — including a machine-readable index — see: - Hub: [adrianpetrasch.com/learnings/](https://adrianpetrasch.com/learnings/) (DE) - Hub: [adrianpetrasch.com/learnings/en/](https://adrianpetrasch.com/learnings/en/) (EN) - LLM index: [adrianpetrasch.com/learnings/llms.txt](https://adrianpetrasch.com/learnings/llms.txt)