Add AI chat integration with byllm for tweet analysis

littleX_BE/littleX.impl.jac → littleX_BE/littleX.impl.jac 20.18.12

@@ -126,4 +126,21 @@ impl load_feed.load {
 impl load_feed.report_feed {
         self.results.sort(key=lambda x:dict:x['similarity'][0], reverse=True);
         report self.results;
+}
+
+impl chat_with_ai.chat {
+    feeds = root spawn load_feed();
+    context = [i.get("Tweet_Info","") for i in feeds.results];
+
+    response = self.generate_response(
+        message=self.message,
+        chat_history=self.chat_history,
+        tweets_context=context
+    );
+
+    self.chat_history.append({"role": "user", "content": self.message});
+    self.chat_history.append({"role": "assistant", "content": response});
+    self.response = response;
+
+    report {"response": response, "chat_history": self.chat_history};
 }

littleX_BE/littleX.jac → littleX_BE/littleX.jac 20.18.12

@@ -2,8 +2,11 @@ import datetime;
 import numpy;
 import from sklearn.feature_extraction.text { TfidfVectorizer }
 import from sklearn.metrics.pairwise { cosine_similarity }
+import from byllm.llm {Model}
 
 glob vectorizer = TfidfVectorizer();
+glob llm = Model(model_name="gemini/gemini-2.5-flash");
+#glob llm = Model(model_name="gemini/gemini-2.5-flash" , verbose=True);
 
 def search_tweets(query: str, tweet:str) -> int;
 
@@ -119,4 +122,17 @@ walker load_feed(visit_profile) {
 
     can report_feed with exit;
 
+}
+
+walker chat_with_ai {
+    has message: str;
+    has chat_history: list[dict] = [];
+    has response: str = "";
+
+    """Generate helpful responses about tweets and social interactions. 
+    Analyze tweet content, provide insights, answer questions about the feed."""
+    def generate_response(message: str, chat_history: list[dict], tweets_context: str) -> str 
+        by llm(method="ReAct", temperature=0.7);
+
+    can chat with `root entry;
 }