%% generate tags start %%
#obsidian
%% generate tags end %%
#obsidian/plugin
## What is It?
%% run start
```ts
const {LinkPreview} = customJS
return LinkPreview.getLinkPreviewFromUrl("https://github.com/brianpetro/obsidian-smart-connections")
```
%%
<div class="nifty-link-card-container">
<a class="nifty-link-card" href="https://github.com/brianpetro/obsidian-smart-connections" target="_blank">
<div class="nifty-link-card-text">
<div class="nifty-link-card-title line-clamp-2">GitHub - brianpetro/obsidian-smart-connections: Chat with your notes in Obsidian! Plus, see what's most relevant in real-time! Interact and stay organized. Powered by OpenAI ChatGPT, GPT-4 & Embeddings.</div>
<div class="nifty-link-card-description">Chat with your notes in Obsidian! Plus, see what&#39;s most relevant in real-time! Interact and stay organized. Powered by OpenAI ChatGPT, GPT-4 &amp; Embeddings. - GitHub - brianpetro/obsi...</div>
<div class="nifty-link-href">
<img class="nifty-link-icon" src="https://github.com/fluidicon.png">
https://github.com/brianpetro/obsidian-smart-connections
</div>
</div>
<div class="nifty-link-image-container">
<div class="nifty-link-image" style="background-image: url('https://repository-images.githubusercontent.com/582432706/e0ab0d8b-3582-470f-bc7e-51f55ee35eb7')">
</div>
</div>
</a>
</div>
%% run end %%
## What I Think They Need to Improve
I think I don't like about the smart connection is that it will create embedding from all your content but some of the content openai has no idea how to handle and openai will train on it as well.
For example the dataview stuff, comment, the html codes. This will not only make you pay more because you use more tokens, but also give inaccurate result.