Advanced techniques

The Importance of Context

The most crucial factor for a high-quality AI response is the context you provide. The more relevant background information, data, and specific details you include in your prompt, the more accurate and useful the result will be.

Think of the AI as a new employee: without the full picture, it can only provide general answers.

Example: Email Response
Instead of asking the AI to respond only to the last email received, include the entire email history in the prompt. This allows the AI to understand the full conversation, the tone, and previous agreements, enabling it to formulate a response that fits perfectly.

Multi-step prompts (step-by-step instructions)

Don’t try to solve complex tasks in a single, massive prompt. A far more effective approach is to break the task down into smaller, logical steps. Engage in a dialogue with the AI, providing feedback after each step and adjusting the direction. This way, you maintain control and ensure the final result aligns exactly with your vision.

Example: Creating a blog post

Prompt 1: "I want to write a blog post on 'The Future of Artificial Intelligence in Marketing.' First, create a detailed outline for me with an introduction, three main points, and a conclusion. Don’t write the text yet."

Prompt 2: "The outline is good. For the first step, just write a captivating introduction that piques the readers’ curiosity."

Prompt 3: "Perfect. Now elaborate on the first main point, 'Personalization through AI,' and provide specific examples."

(This process continues until the entire article is complete.)

This approach has two key advantages: First, it’s much easier for you as a user to review small increments and provide targeted feedback, allowing the AI to develop a more precise understanding of the end goal over time.

Second, this method conserves the context window, saves tokens, and thus minimizes the risk of “hallucinations”—the invention of facts during overly long dialogues.

Assistant Prompts (Persona Pattern)

This is one of the most powerful techniques for transforming the AI into a specialized assistant for recurring tasks. Instead of developing a one-off strategy, you assign the AI a precise role and provide it with a repeatable, clearly structured task.

The primary goal is to achieve reliable and consistent results with every single application. By providing the AI with clear guidelines in the form of a role, objective, format, and constraints, you eliminate ambiguity and prevent it from delivering generic or unwanted responses.

This approach is ideal for automating routine tasks, as it ensures that the quality remains exactly the same with every execution.

The structure of such a prompt looks like this:

• Role: Who is the AI supposed to be? (e.g., a nutrition expert, an SAP consultant for small and medium-sized businesses)

• Task/Goal: What is the primary goal of the prompt? What task needs to be completed?

• (optional) Workflow/Step-by-Step Instructions: A detailed process for the AI to follow. See example below.

• Context: All necessary background information required to complete this task.

• Output/Format: Specific guidelines for the structure of the output.

• Limitations: Restrictions and prohibitions (e.g., tone, word count)

• (optional) Examples: Samples of the desired result/output.

Example 1: Create social media posts from blog articles

Role: You are an experienced content marketing assistant specializing in repurposing blog articles for various social media channels.
Task/Goal: Your recurring task is to create three different social media posts from an existing blog article to maximize its reach.
Context: Here is the text of the blog article to be promoted:[Blogartikel-Text hier einfügen]
Output/Format: Create three posts: 1. A LinkedIn post (approx. 150 words, professional tone, with 3–4 relevant hashtags), 2. An X (Twitter) thread (3 tweets, starting with a hook), 3. An Instagram caption (more emotional, with a question at the end and 5–7 hashtags).
Restriction: Do not use clickbait phrasing. The posts must accurately reflect the content of the article and must not add any information not contained in the original text.

Example 2 (with workflow): Handling customer inquiries

Role: You are an experienced first-level support agent. Your strength lies in quickly categorizing customer inquiries and taking the right initial steps.
Task/Goal: Your daily task is to categorize new customer inquiries and draft an initial response.
Workflow/Step-by-Step Instructions:
1. Analyze the inquiry and assign it to one of the following categories: Technische Frage, Rechnungsfrage, Feature-Wunsch.
2. Pause: First, present only the category to me and wait for my confirmation.
3. After my confirmation, create a draft response based on the category and the appropriate example.
Context: Here is the new customer inquiry:[Kundenanfrage hier einfügen]
Output/Format: The final response should be formatted as email text that can be copied directly. Always start with "Hi [Customer Name],".
Limitation: The response must be empathetic but standardized. Do not provide specific timelines for bug fixes or new features.