#005 AI | PromptWizard Automating Prompts for LLMs

Conway’s Law, a principle suggesting that the technical architecture of a software system inevitably reflects the communication patterns of the organization that built it.

Vertical Slice Architecture (VSA) as an alternative to traditional layered architecture, emphasizing feature-based organization for improved maintainability. It then explains how the Cortex.Mediator library, part of the Cortex Data Framework, facilitates VSA implementation in .NET applications by employing the Mediator pattern with commands, queries, and notifications within each slice.

PromptWizard is an automated framework for optimizing prompts for large language models (LLMs). This system uses a self-evolving mechanism that iteratively refines prompts and examples through a feedback loop of generation, critique, and synthesis. PromptWizard aims to overcome the limitations of manual prompt engineering by improving task performance across diverse applications while also demonstrating cost-effectiveness compared to existing optimization strategies. The research demonstrates PromptWizard's effectiveness even with limited data and smaller LLMs, highlighting its practical utility.

In our 4th episode, we will discuss on how to build reactive architectures by transitioning from monolithic systems to distributed microservices. It uses a payment processing example to illustrate the shift from synchronous, pull-based interactions to asynchronous, push-based event-driven communication. The benefits of this transition include improved scalability, resilience, and service autonomy. Key concepts discussed are microservices, reactive principles (responsiveness, resilience, elasticity, message-driven), and the advantages of a push model over a pull model.

At first, designing microservices may seem simple. However, as the system and teams expand, and dependencies increase, poor design choices can have major sociotechnical consequences. It becomes challenging to correct these issues, as they are built upon sub-optimal design layers. To avoid these problems, it is crucial to consider how consumers impact the design of microservices. Dependencies and misplaced responsibilities can be easily introduced, costly to endure, and difficult to reverse. Understanding the fundamental patterns, such as facade contexts, and BFFs, can also help in designing microservices effectively.

This is the second episode of a series explaining the Cortex Data Framework, focusing on its newly implemented Change Data Capture (CDC) functionality. The episode highlights Cortex's expanded support for various databases, including Microsoft SQL Server, PostgreSQL, and MongoDB. The primary goal is demonstrating how developers can leverage CDC within Cortex Streams to respond to database changes and implement custom logic. Essentially, it's a tutorial explaining how to use a new feature to build more reactive applications.

In our first episode of Inside 2025, we are focusing on the Cortex Data Framework. Stay tuned as we analyze the documentation of Cortex. Get ready to start streaming! 🚀🚀🚀