AI Course for "Language Models"
LLMs for Text, Research, and Code
Learn how modern AI systems like ChatGPT, Gemini, Claude, and Grok work, what their limitations are, and how they can support text, code, research, automation, and business processes.
Your AI Expert
Johann Dirschl
Three AI Courses for a Practical Start
ChatGPT, Gemini, Grok, Claude, and other language models: Prompts, limitations, costs, data privacy, API vs. Chat – for everyday use, business, and development.
Coming Soon 2. AI for Images, Speech, Music, and Video
To the Generative Media Course - suitable tools, law, marketing, photography, and web projects.
Coming Soon 3. AI in Business
Course for AI in Business - Automation, API, RAG, local LLMs, corporate solutions, phone assistants, cost control. Target Audience: Families, self-employed individuals, companies, associations, creatives, and developers – without empty buzzwords, with real application.
LLMs – More Than Just a Better Chat
Large Language Models (LLMs) understand and generate language – often including program code. They act like chatbots, but are a text tool, research aid, learning partner, programming assistant, and structurer all in one.
Typical Applications
- Improve texts and emails, prepare blog posts, product descriptions
- Technical questions, WordPress, HTML, CSS, JavaScript, databases
- Translate, brainstorm ideas, keywords, debugging, and explanations
Why LLMs are More Than Just a Better Google Search
A search engine provides links. An LLM provides answers, explanations, and solutions – you can ask follow-up questions, request simplification, or improve step-by-step. A dialogue emerges, not just a list of results.
Important: AI is not infallible truth. Always verify information regarding law, medicine, taxes, security, or business decisions. Good workflow = ask, verify, improve, apply.
Why Can Incorrect Answers Occur?
LLMs calculate probable word sequences – they don't automatically "know" if a statement is true. If information is missing or training and the question are unclear, a convincingly sounding but incorrect answer can arise.
In the course, you will learn control through better prompts, clear tasks, source verification, counter-questions, roles, limitations, and plausibility checks.
How Detailed Does a Prompt Need to Be?
A prompt doesn't have to be long – but it must be clear. For simple tasks, a single sentence is often enough. For important results, context, goal, target audience, style, and format help.
When is a Chatbot Enough – When Do You Need an API?
| Situation | Chatbot | API |
|---|---|---|
| Typical Use | Individual texts, ideas, translations, explanations, code assistance | Automatic integration into website, plugin, app, CRM, support |
| Examples | Manually in ChatGPT, Gemini, Claude, Grok | FAQ assistant, product texts, translations, document analysis, phone bots |
| Short Rule | For manual use, chat is sufficient. | For automation, an API is usually needed. |
What Does AI Cost?
A chat subscription is predictable. APIs often charge by tokens (text units) – the longer the input and response and the more powerful the model, the higher the costs can be.
| Model | Typical Billing | Note |
|---|---|---|
| Chat Subscription | Monthly per user | Often sufficient for private & small teams |
| API | By usage (tokens) | Input + response + model strength determine the price |
| Additional | Image, audio, video, storage, automation | For businesses, plan costs |
Where Does My Data Go?
This depends on the provider, tariff, and technical implementation. Free chat, business plan, enterprise, API, and local model are different scenarios.
For customer data, contracts, internal documents, or trade secrets: consciously choose the provider, data privacy settings, and, if necessary, a local or corporate solution.
Local LLMs, Corporate LLMs, and RAG
Run on your own PC or server – data does not necessarily have to go to the cloud. Require computing power; small models for summaries and internal tests, large cloud models often still stronger in complexity.
AI for businesses: Rights, logging, teams, integration – when many employees and internal documents are involved.
Retrieval-Augmented Generation: AI uses its own PDFs, manuals, websites – ideal for support, training, and knowledge management without training everything into the model.
AI Assistants and Phone Assistants
Systems can listen, speak, and be connected – from speech recognition, LLM, data source, speech output, and clear rules for handover and data privacy.
Overview of Known Systems
| System | Strengths / Typical Use |
|---|---|
| ChatGPT | All-round: Text, code, ideas, analysis, creative workflows |
| Claude (e.g., Sonnet) | Long texts, structured analysis, programming, clean language |
| Gemini | Google ecosystem, multimodal, large contexts |
| Grok | Dialogue, current topics, creative applications (xAI) |
| Perplexity | Research with source attribution |
| ElevenLabs | AI voices, voiceover, audio |
| GitHub Copilot | Code directly in the IDE |
| Ollama / LM Studio | Local models on your own computer |
The course is not about a single "best" tool, but about choosing the right tool for the task.
Who is this Course Suitable For?
- Private & Family: Writing, learning, researching, planning, creative work
- Businesses: Acceleration, data privacy, initial automation, meaningful processes
- Development: Code, APIs, WordPress, WebApps, prototyping
Course Goal
At the end, there's no vague "AI can write texts." You will understand: What can AI do? What can't it do? Which tool for what purpose? How to write better prompts? When to use API or local? How to protect data?
We share practical knowledge from text work, web development, WordPress, automation, AI music, image generation, SEO, and our own projects – parts of our websites and plugins were also created with AI support.
Book AI Course or Company Training
Questions about dates, content, or a program just for your team? Get in touch – we will tailor the course to your use cases.