Add podcasts

Author: mythz

MyApp/_podcasts/2025-10-28_ai-chat.md

@@ -0,0 +1,21 @@
+---
+title: AI Chat - A Simple OpenAI Chat Completions API, UI & Client LLM Gateway
+summary: Unlock the value of OpenAI Chat APIs using a simple, serializable ChatCompletion DTO that works everywhere
+tags: [llms,ai,openai,chat]
+author: Demis Bellot
+url: https://media.servicestack.com/podcasts/apikey_auth_apis.mp3
+media: {size:13193573,duration:950.256327,format:mp3}
+---
+
+This episode introduces **AI Chat**, a new solution designed to simplify the integration of AI into applications 
+by focusing entirely on the **OpenAI Chat Completion API**. This approach prioritizes a simple, serializable
+`ChatCompletion` **Data Transfer Object (DTO)** which acts as the core abstraction, contrasting sharply 
+with overly complex frameworks like **Microsoft Semantic Kernel**. 
+
+It's argues that heavyweight, non-portable abstractions quickly become irrelevant as AI models advance, 
+leading to vendor lock-in and required rewrites. By centering the design around the `ChatCompletion` DTO, 
+**ServiceStack.AI.Chat** offers advantages like easy data persistence, client-side usability, and 
+simplified debugging, while also supporting multi-modal content like images and audio. 
+
+The solution includes robust **model routing and failover** capabilities configured via a simple `llms.json` file, 
+and is compatible with its Python counterpart, [llms.py](https://github.com/ServiceStack/llms).

MyApp/_podcasts/2025-10-31_ai-chat-ui.md

@@ -0,0 +1,21 @@
+---
+title: AI Chat - A Customizable, Private, ChatGPT-like UI
+summary: Unlock the value of OpenAI Chat APIs using a simple, serializable ChatCompletion DTO that works everywhere
+tags: [llms,ai,openai,chat]
+author: Demis Bellot
+url: https://media.servicestack.com/podcasts/apikey_auth_apis.mp3
+media: {size:10659555,duration:756.897959,format:mp3}
+---
+
+This episode details the features and implementation of **AI Chat**, a customizable, private, ChatGPT-like user 
+interface built on **ServiceStack.AI.Chat** for .NET 8+ projects. A core value proposition is enabling 
+organizations to offer a controlled AI experience while maintaining **Fast, Local, and Private** access, 
+managing their own API keys and billing. 
+
+Users can interact with the AI through a **Simple and Flexible UI** that supports **Rich, Multimodal Inputs** 
+including image, audio, and file attachments, going beyond simple text conversations. 
+
+Access is secured using **Identity Auth or Valid API Key** mechanisms, and the interface offers advanced 
+customization for appearance, default system prompts, and **Custom AI Chat Requests** for fine-tuning model 
+parameters. AI Chat also includes comprehensive features like a **Comprehensive System Prompt Library**, 
+**Search History**, and tools for **Admin Users** to manage and prioritize different AI providers.

MyApp/_podcasts/2025-11-04_ai-chat-explorer.md

@@ -0,0 +1,21 @@
+---
+title: Creating a custom Explorer UI for OpenAIs Chat API
+summary: Learning how to create a customized API Explorer API UIs like ChatCompletion's UI
+tags: [llms,ai,openai,chat]
+author: Demis Bellot
+url: https://media.servicestack.com/podcasts/apikey_auth_apis.mp3
+media: {size:11945780,duration:837.328980,format:mp3}
+---
+
+The episode discusses the process of **creating a custom API Explorer UI** specifically for OpenAI's Chat API, 
+contrasting it with generic interfaces like Swagger UI. It asserts that **generic UIs fall short** when dealing 
+with complex APIs, offering minimal assistance and poor usability, while ServiceStack's API Explorer provides a 
+more functional, **hand-crafted application UI experience**. 
+
+The key to this improved usability lies in **optimizing UI components** for different data types—such as using 
+number inputs for numeric fields and Comboboxes for options like **Model** and **Service Tier**—along with 
+providing concise **help text hints** and client-side validation. 
+
+Furthermore, the episode explains how custom components, like the specialized **ChatMessages component**, can be 
+integrated to handle complex properties, all driven by declarative **[Input]** and **[FieldCss]** attributes on 
+the Request Data Transfer Object (DTO).