---

Last year I wrote a basic blog post about using AI in software development (https://rocketsoftware.tech/blog/post/using-ai-to-help-with-code-writing). Over the last few months, I took on a project that would alter how I used AI for my projects.

In years past, I would always use Google to help me find answers to my questions when it came to how perform a specific task in C#, which was always effective but was also much more time consuming. For this project, I used Google's Gemini and my locally hosted Ollama server as my assistants. Both assistants were considerably more helpful than in years past. The project took just under 500 hours from conception to a fairly usable product (the product is SynapseKeep, a AI enhanced personal knowledge base). I build the project using .Net Aspire, which was fairly new to me.

Writing code isn't the hardest part, it's writing efficient code that's the hardest. When you're integrating with some many different third-party products (like external AI engines such as Gemini or Ollama Server and using a tool like Searxng and Firecrawler to get outside information), code execution efficiency is paramount, and that's where the assistants helped me the most, showing me best practices, helping set the right context windows, making sure my embedding vector sizes were right for my environment, all of that matters. The math involved on vector sizes and context windows alone is crazy.

The biggest thing that I learned throughout this project is that I still need to be able to piece it all together, AI still misses on some of the integration needs (outdated endpoints, etc.), and building and testing is even more critical than ever before.

---

What is Synapse?

I recently started using new personal knowledgebase programs that I thought would be good for what I needed to do, but I found I wanted something more like NotebookLM, but while that's great for research, it's not really a PKB at this point.  So, I came up with an idea to see if I could create my own.  I know that tools like Obsidian have this level of web/AI integration already, but I don't like having to use things like Syncthing or some other file synching system.   Here is the basic goal:

My goal of this project is to make my own personal knowledge base that can also give me results from the web that can help augment or provide insights into my notes. Like NotebookLM but also like Obsidian notes. I've got the searching for my notes, but I want to expand that into web information.

#C-Sharp #C# #AI #ChatGPT

I recently started using ChatGPT and my own local Ollama server.  The code is generally pretty good, and requires some tweaking, as you'd suspect.  For example, here is a code snippet of ChatGPT trying to show how to perform Name Entity Recognition.

using Microsoft.SemanticKernel;
using Microsoft.SemanticKernel.ChatCompletion;
#pragma warning disable SKEXP0070

// Create two kernels for each model
var intentKernel = Kernel.CreateBuilder()
                   .AddOllamaChatCompletion(
                       modelId: "llama3.2:latest",
                       endpoint: new Uri("http://192.168.2.19:11434"))
                   .Build();

var nerKernel = Kernel.CreateBuilder()
                   .AddOllamaChatCompletion(
                       modelId: "ner-model:latest",
                       endpoint: new Uri("http://192.168.2.19:11434"))
                   .Build();