19 lines
1.3 KiB
Plaintext
19 lines
1.3 KiB
Plaintext
<p>Recently I stumbled upon <a href="https://github.com/semiosis/pen.el">pen.el</a> which interacts with language models, like OpenAI GTP-3.</p>
|
||
|
||
<p>This is interesting. Loosely thinking, language models are good complement to Emacs as both are operating on plain text.</p>
|
||
|
||
<p>I tried pen.el on a Docker container, but it didn’t work out of box at my first try.</p>
|
||
|
||
<p>Also, I look at the <a href="https://beta.openai.com/docs/api-reference/introduction">OpenAI API documentation</a> and it doesn’t seem hard to build a wrapper around it.</p>
|
||
|
||
<p>After a weekend’s work, here is the first implementation of <a href="https://github.com/junjizhi/aide.el">aide.el</a>.</p>
|
||
|
||
<blockquote>
|
||
<p>Fun fact: The repo name <strong>aide</strong> is generated by <a href="https://beta.openai.com/examples/default-product-name-gen">OpenAI Product Name Generator</a>, which reads like AI + IDE, or AI Dwells (in) Emacs :P</p>
|
||
</blockquote>
|
||
|
||
<p>Think of aide.el as a thin wrapper around APIs like OpenAI.</p>
|
||
|
||
<p>Currently <code class="language-plaintext highlighter-rouge">aide.el</code> implementation is minimal: You select a region of text in Emacs, run <code class="language-plaintext highlighter-rouge">M-x aide-openai-completion</code>, and Emacs shows how AI responds to it.</p>
|
||
|
||
<p><a href="https://github.com/junjizhi/aide.el">Try it out</a> and let me know what you think!</p> |