Posts
-
MCP - The USB-C for AI
Large language models (LLMs) like GPT-4 or Claude are incredibly powerful, but they have a big limitation: by default they only know what’s in their training data and what you prompt them with. They don’t automatically have access to your live data, files, or tools. This means if you want an AI assistant to work with your documents, databases, or APIs, you traditionally have to write custom code or plugins for each case. That approach is clunky and hard to scale – every new data source or tool requires yet another integration. Enter the Model Context Protocol (MCP), an open standard that aims to simplify and standardize how LLMs connect to external data and services . Think of MCP as a way to give AI assistants a universal “plug” into the information and tools they need, no matter where those live.
-
LLM Evaluation with Prompt Flow SDK
In my previous blogpost I discussed how you could do LLM evaluation with RAGAS. In this one we will keep the structure and try to do exactly the same but with Azure PromptFlow SDK
-
LLM Evaluation with RAGAS
Embarking on the journey of evaluating your RAG LLM application can be both exciting and daunting. It’s a process that not only tests the capabilities of the model but also reflects your own understanding of its intricate workings. In this blog post, we’ll dive into the key aspects of evaluation, from setting up the right metrics to interpreting the results for continuous improvement. Whether you’re a seasoned data scientist or a curious enthusiast, this guide aims to shed light on the nuances of machine learning evaluation and its significance in the ever-evolving field of AI.
-
Azure Function with FastAPI + Non HttpTriggers
Integrating FastAPI with Azure Functions can enhance the functionality of serverless applications, especially when you’re looking to create robust APIs with Python. While FastAPI doesn’t natively support Azure-specific triggers like Queue, Timer, or Blob, it’s possible to set up a workaround. This involves using the Azure Functions’ bindings and connecting them with FastAPI endpoints. By doing so, you can enjoy the benefits of FastAPI’s features, such as automatic Swagger documentation and data validation, while also leveraging Azure Functions’ powerful event-driven triggers.
-
Query Fields in Document Intelligence: A Hidden Gem
Query fields in Document Intelligence are a powerful feature that can significantly enhance your document processing capabilities. This feature allows you to extract specific fields from documents without the need to train a custom model. Instead, you can define the fields you want to extract, and the model will only extract the corresponding values. This is particularly useful when the values you need cannot be described as key-value pairs in the document, such as the agreement date of a contract.
-
Reciprocal Rank Fusion (RRF), Euhm what?
Yes you heard me well, I said Reciprocal Rank Fusion. It is a mouth full, but such an important method to improve your search results when using RAG. In below article I will explain what it is and how you can make use of it when using Azure AI Search.
-
Sending images to OpenAI via the C# SDK
While preparing a training for a customer, one of the examples I want to show was how you can send images to the OpenAI ChatCompletion endpoint. I could not find any examples online how to do it in C# so I delft into the sourcecode, try to understand how it should be done.
subscribe via RSS