<!DOCTYPE html>
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<p><font size="4">Indeed, this RAG things looks interesting in our
context of Smalltalk. </font></p>
<p><font size="4">I experienced that for specific domain modeled in
Smalltalk, injecting "by hand", in a prompt, DSL knowledge gives
great results. Particularly for non programmer that you want to
acquaint to programming.</font></p>
<p><font size="4">Hilaire </font></p>
<div class="moz-cite-prefix">Le 11/01/2026 à 03:35, Phil B via
Cuis-dev a écrit :<br>
</div>
<blockquote type="cite"
cite="mid:CAMJMOehgtQsL9hkf9bwpDTeAoL-N3An54GcRGSDFV-EvEQR+QA@mail.gmail.com">
<div dir="ltr">Current LLMs aren't that smart. They're purely
tokens in, tokens out with no real structural understanding.
The idea of RAG is to inject the LLM with the knowledge needed
for a given prompt to give a good answer without additional
training required.</div>
</blockquote>
<pre class="moz-signature" cols="72">--
<a class="moz-txt-link-freetext" href="http://mamot.fr/@drgeo">http://mamot.fr/@drgeo</a></pre>
</body>
</html>