Level 3 — Pro
You want to get the most out of it. This level isn't just about better prompts anymore — here you'll learn how to integrate AI into your workflows, automate processes, and let AI agents work for you. Welcome to the pro level.
No-Code Automation: n8n and Make
Imagine receiving an email, and a system automatically reads it, categorises it, drafts a reply, and logs everything in a spreadsheet — without you lifting a finger. That's exactly what automation platforms enable.
n8n
Open-source automation that you can run on your own server. Especially popular in the AI community because of its strong AI integration. You connect different “nodes” visually — like building blocks in a flowchart.
Make (formerly Integromat)
Cloud-based platform with over 1,500 app integrations. More beginner-friendly than n8n, but less flexible for complex AI workflows. Free tier available.
Example Workflow: Automatic Email Response
Trigger: New Email
n8n monitors your inbox. As soon as a new email arrives, the workflow starts.
AI Analyses the Content
The email is sent to ChatGPT/Claude: “Categorise this email (enquiry, complaint, spam) and create a draft reply.”
Save the Result
Category and draft are saved in Google Sheets. For urgent emails, you get a Slack notification.
AI Agents: What Are They?
Until now, you've used AI to answer questions. AI agents go a step further: they can independently take action. An agent receives a goal, devises a plan, executes the individual steps, and adapts its strategy if something doesn't work.
Chatbot vs Agent — The Difference
Chatbot: You ask, it answers. Done.
Agent: You give a goal, it researches, plans, executes, checks the result, and reports back.
As of 2026, AI agents are at an early but usable stage. Examples:
- Computer Use: AI agents that can operate your browser or computer
- Research Agents: They search the internet, read pages, and create reports
- Coding Agents: They write, test, and improve code independently
- Workflow Agents: They execute multi-step tasks in n8n or Make
Important:Agents are powerful but not perfect. Always review their results — especially for tasks with real-world consequences.
Understanding APIs (Without Being a Programmer)
The word “API” sounds technical, but it's simpler than you think. An API (Application Programming Interface) is like a waiter in a restaurant: you (the app) place your order, the waiter (the API) takes it to the kitchen (the server), and comes back with the result.
Why Is This Relevant for You?
- Automation tools like n8n and Make use APIs to connect tools together
- Via APIs, you can process hundreds of texts simultaneously instead of typing them one by one
- You can build AI features into your own projects (websites, apps)
What Is an API Key?
An API key is like a personal ID card for accessing a service. You get one when you register with a provider (e.g., OpenAI, Anthropic). Treat your API key like a password — don't share it publicly.
To start, you don't need a single line of code. Tools like n8n, Make, or “GPT for Sheets” handle the API communication for you. If you want to go deeper, AI tools themselves can help you write simple scripts.
Building Your Own Chatbots
You can build an AI assistant that answers exactly the way you want — with its own knowledge, style, and rules. And completely without programming.
Choose a Platform
OpenAI GPT Builder: Create custom GPTs directly in ChatGPT. Claude Projects: Set up a project with system prompt and knowledge documents.
Define the Personality
Write a system prompt that establishes tone, style, boundaries, and behaviour.
Feed It Knowledge
Upload PDFs, FAQs, product info, or guides. The chatbot uses this knowledge in its responses.
Test and Refine
Ask typical questions and check whether the answers are correct. Adjust the system prompt until the result fits.
You are the customer support assistant for [Company Name]. Respond in a friendly, concise, and helpful manner. Only use information from the uploaded documents. If you can't answer a question, say: “I'd be happy to connect you with our team — email us at support@company.com.” Never make up information.
Prompt Chains and Multi-Step Workflows
Individual prompts are good. But the real magic happens when you connect prompts into chains — each step builds on the result of the previous one.
Example: Content Pipeline for a Blog Article
Keyword Research: “Give me 10 SEO keywords on the topic of [X] for the English market.”
Outline: “Create an outline for a 1,500-word article based on the top 3 keywords.”
Writing: “Write the article based on this outline. Tone: casual-informative.”
Review: “Check the article for facts, redundancies, and SEO optimisation.”
Social Media: “Create 3 social media posts (LinkedIn, Instagram, X) based on the article.”
In n8n or Make, you can fully automate such chains. But even manually in the chat, this step-by-step approach is far more effective than a single massive prompt.
Using AI in Teams
AI becomes even more powerful when a whole team uses it. The key: shared standards so everyone gets the same quality.
Shared Prompt Libraries
Create a shared document with your best prompts. Categorise by task: emails, social media, reports, customer service. Everyone on the team uses the same tested templates.
SOPs with AI Support
Standard Operating Procedures (SOPs) define how certain tasks are done. Add AI prompts directly to your SOPs: “Step 3: Use the following prompt to summarise the monthly report: [Prompt]”
Custom Team GPTs / Projects
Build specialised chatbots for recurring tasks: one for onboarding new employees, one for customer support, one for content creation. Everyone on the team has instant access to the collected knowledge.
Practical Tip
As a team, define clear rules: What may be entered into AI tools, what may not? Who reviews the results? How are AI-generated contents labelled?
Ethics and Responsibility as a Pro User
With great power comes great responsibility. As an advanced AI user, you should be aware of these points:
When NOT to Use AI
- When the decision has irreversible consequences (law, medicine, finance) — without human review
- When you can't assess the results yourself
- When personal data is involved and the privacy situation is unclear
Bias Awareness
AI models reflect the biases of their training data. When using AI to evaluate applications, write texts about people, or make recommendations, check the results for fairness and balance.
Transparency and Disclosure
When AI has substantially contributed to creating content, communicate this openly. Especially with: publications, customer communications, and educational materials. Honesty builds trust.
The AI landscape is changing rapidly. Stay informed about current developments, best practices, and regulations. You can find more on this in our AI course.
Ready for more?
You've reached the pro level — automation, APIs, and AI agents are now part of your toolkit. But there are two more levels to go: in Level 4 you'll build your own AI systems with RAG and MCP, and in Level 5you'll become an AI architect.