2025
Sep
18
LLM (Large Language Model) Top-P works
Use Cases Low Top-P value Use Case: Customer Support Why? High Top-P value Use Case: Creative writer Why?
Tag: agent
2025
Sep
18
Cost and value balance with LLMs (LLM parameters – LLM Top-P)
The LLM parameter Top-P, also known as nucleus sampling, controls the diversity of the output by setting a cumulative probability threshold for selecting the next token. It is used to produce higher quality and more diverse outputs depending on the setting. When generating text, tokens (words, sentences
2025
Sep
18
MCP (Model Context Protocol)
With MCP Servers you have the capability to integrate a bunch of tools and capabilities into your AI / LLM you prefer. That means you can integrate with several services you want like email-systems, CRM-systems or databases for instance. What that exactly
2025
Sep
18
Cost and value balance with LLMs (LLM parameters – LLM temperature)
With this setting you can choose the right balance between randomness and determinism of the outcome generated by the LLM. An important aspect in applications where decisions should be made based on profound facts. The other side, if you need more creativity
2025
Sep
18
Choosing an LLM model
To get started with AI and agents the very first step is to choose an LLM you want to work with, OpenAI, Gemini Anthropic or which ever you choose. Here are some aspects you should take into consideration for making your choice.
2025
Sep
18
Cost and value balance with LLMs (LLM parameters – Max tokens)
With LLM parameters you have the chance to configure additional important settings for your LLM. With that settings you can influence the balance of costs and value for instance. But in addition, you can influence how the output will be generated, is
2025
Sep
18
AI Prompt Engineering
If you working in the IT space of Agentic AI and automation, one skill is very important, PromptEngineering! After I did the course #ChatGPT Prompt Engineering for Developers by @OpenAI and @DeepLearningAI, last week I did some courses to optimize and increase