diff --git a/modules/features/teamsbot/service.py b/modules/features/teamsbot/service.py
index 01f84ae7..e481bf86 100644
--- a/modules/features/teamsbot/service.py
+++ b/modules/features/teamsbot/service.py
@@ -565,6 +565,7 @@ class TeamsbotService:
             if (
                 source == "speakerHint"
                 and isFinal
+                and not self._isBotSpeaker(speaker)
                 and self.config.responseMode != TeamsbotResponseMode.TRANSCRIBE_ONLY
             ):
                 shouldTriggerFromHint = self._shouldTriggerAnalysis(text, allowPeriodic=False)
@@ -587,8 +588,9 @@ class TeamsbotService:
             return
 
         # Chat history: messages sent before the bot joined the meeting.
-        # Stored in DB and context (available if someone refers to chat history)
-        # but never used to trigger AI responses.
+        # Stored in DB for reference but NOT added to the AI context buffer,
+        # because old messages (e.g. "nyla, summarize the protocol") would
+        # be treated as current requests when AI analysis is triggered.
         if source == "chatHistory":
             transcriptData = TeamsbotTranscript(
                 sessionId=sessionId,
@@ -601,16 +603,6 @@ class TeamsbotService:
             ).model_dump()
             createdTranscript = interface.createTranscript(transcriptData)
 
-            self._contextBuffer.append({
-                "speaker": speaker or "Unknown",
-                "text": text,
-                "timestamp": getUtcTimestamp(),
-                "source": "chatHistory",
-            })
-            maxSegments = self.config.contextWindowSegments
-            if len(self._contextBuffer) > maxSegments:
-                self._contextBuffer = self._contextBuffer[-maxSegments:]
-
             await _emitSessionEvent(sessionId, "transcript", {
                 "id": createdTranscript.get("id"),
                 "speaker": speaker,