diff --git a/backend/app/api/routers/ws.py b/backend/app/api/routers/ws.py
index f9f5519..f659b54 100644
--- a/backend/app/api/routers/ws.py
+++ b/backend/app/api/routers/ws.py
@@ -137,7 +137,7 @@ def cut_fragment_video(self, path: str, start_ms: int, end_ms: int) -> bool:
 
     def extract_audio_from_video(self, video_path: str) -> str | None:
         """Extract audio from video file and save as WAV with same base filename."""
-        # Generate audio file path with .wav extension
+        # Generate audio file path with .wav extensionu
         audio_path = video_path.replace(".webm", ".wav")
 
         logger.debug("Extracting audio from %s to %s", video_path, audio_path)
@@ -439,36 +439,6 @@ async def websocket_video_stream(
                 if text_data == "audio-ready":
                     await service.handle_audio_marker()
                     continue
-                if text_data is not None:
-                    # Treat any other text frame as a direct user answer (debug mode support)
-                    logger.info(
-                        "Received direct text on WS for interview %s: %s",
-                        interview_id,
-                        text_data[:200],
-                    )
-                    await service.update_state(InterviewSocketState.GENERATING_RESPONSE)
-                    latest_message = await service.submit_user_answer(
-                        interview_id, text_data.strip()
-                    )
-                    if latest_message is None:
-                        logger.warning(
-                            "Failed to submit direct text for interview %s",
-                            interview_id,
-                        )
-                    else:
-                        logger.info(
-                            "Direct text submitted successfully for interview %s",
-                            interview_id,
-                        )
-                        if settings.use_yandex_speech_synthesis:
-                            await service.update_state(
-                                InterviewSocketState.SPEECH_SYNTHESIS
-                            )
-                            await service.send_message_to_user(latest_message)
-                    await service.update_state(
-                        InterviewSocketState.AWAITING_USER_ANSWER
-                    )
-                    continue
 
             if data_bytes and len(data_bytes) > 0:
                 service.add_video_chunk(data_bytes)
diff --git a/backend/app/services/pdf_parser.py b/backend/app/services/pdf_parser.py
index 27f9250..7a1ab30 100644
--- a/backend/app/services/pdf_parser.py
+++ b/backend/app/services/pdf_parser.py
@@ -158,7 +158,7 @@ async def parse_cv(
             # Generate JSON schema for precise parsing
             json_schema = self._get_cv_json_schema()
 
-            result = self.gigachat_client.chat(
+            result = await self.gigachat_client.achat(
                 {
                     "messages": [
                         {
@@ -339,7 +339,7 @@ async def parse_vacancy(
             # Generate JSON schema for precise parsing
             json_schema = self._get_vacancy_json_schema()
 
-            result = self.gigachat_client.chat(
+            result = await self.gigachat_client.achat(
                 {
                     "messages": [
                         {
@@ -472,7 +472,7 @@ async def analyze_cv_with_gigachat(self, file_id: str, question: str) -> str:
             PDFParsingError: If analysis fails
         """
         try:
-            result = self.gigachat_client.chat(
+            result = await self.gigachat_client.achat(
                 {
                     "function_call": "auto",
                     "messages": [
@@ -502,7 +502,7 @@ async def analyze_vacancy_with_gigachat(self, file_id: str, question: str) -> st
             PDFParsingError: If analysis fails
         """
         try:
-            result = self.gigachat_client.chat(
+            result = await self.gigachat_client.achat(
                 {
                     "function_call": "auto",
                     "messages": [
diff --git a/frontend/nginx.conf b/frontend/nginx.conf
index 6c25588..99ab337 100644
--- a/frontend/nginx.conf
+++ b/frontend/nginx.conf
@@ -74,6 +74,23 @@ http {
             }
         }
 
+        # WebSocket proxy for /ws/ endpoints
+        location /ws/ {
+            proxy_pass http://backend:8000;
+            proxy_http_version 1.1;
+            proxy_set_header Upgrade $http_upgrade;
+            proxy_set_header Connection "upgrade";
+            proxy_set_header Host $host;
+            proxy_set_header X-Real-IP $remote_addr;
+            proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
+            proxy_set_header X-Forwarded-Proto $scheme;
+            
+            # WebSocket specific timeouts
+            proxy_read_timeout 86400;
+            proxy_send_timeout 86400;
+            proxy_connect_timeout 60;
+        }
+
         # Proxy regular API requests to backend
         location /api/ {
             proxy_pass http://backend:8000/;
diff --git a/frontend/src/components/interviews/interview-chat.tsx b/frontend/src/components/interviews/interview-chat.tsx
index 7e5e80a..fdab759 100644
--- a/frontend/src/components/interviews/interview-chat.tsx
+++ b/frontend/src/components/interviews/interview-chat.tsx
@@ -32,10 +32,12 @@ export function InterviewChat({ interviewId }: Props) {
 		refetchInterval: 1000,
 	});
 	const isFinished = interview.data.state === "done";
-	const { webcamRef, sendAudioReadyMarker, socketState, sendTextMessage } =
-		useWebcamStreaming(interviewId, {
+	const { webcamRef, sendAudioReadyMarker, socketState } = useWebcamStreaming(
+		interviewId,
+		{
 			disabled: isFinished,
-		});
+		},
+	);
 	const candidate = useSuspenseQuery(
 		candidateQueryOptions(interview.data.candidate_id),
 	);
@@ -170,7 +172,6 @@ export function InterviewChat({ interviewId }: Props) {
 													e.key === "Enter" &&
 													debugPrompt.trim().length > 0
 												) {
-													sendTextMessage(debugPrompt.trim());
 													setDebugPrompt("");
 												}
 											}}
@@ -178,7 +179,6 @@ export function InterviewChat({ interviewId }: Props) {
 										<Button
 											onClick={() => {
 												if (debugPrompt.trim().length === 0) return;
-												sendTextMessage(debugPrompt.trim());
 												setDebugPrompt("");
 											}}
 										>
diff --git a/frontend/src/hooks/use-webcam-streaming.ts b/frontend/src/hooks/use-webcam-streaming.ts
index a6ea08c..ec2fac7 100644
--- a/frontend/src/hooks/use-webcam-streaming.ts
+++ b/frontend/src/hooks/use-webcam-streaming.ts
@@ -212,16 +212,6 @@ export function useWebcamStreaming(
 		setSocketState("speech_recognition");
 	}
 
-	/**
-	 * Отправка текстового сообщения напрямую по WebSocket (dev/debug режим)
-	 */
-	function sendTextMessage(text: string) {
-		if (!text) return;
-		console.log("Sending text message over WS:", text);
-		sendMessage(text);
-		setSocketState("generating_response");
-	}
-
 	function startRecording() {
 		console.log("Attempting to start recording...");
 		setIsRecording(true);
@@ -258,7 +248,6 @@ export function useWebcamStreaming(
 		startRecording,
 		handleStopRecording,
 		sendAudioReadyMarker,
-		sendTextMessage,
 		socketState,
 	};
 }