diff --git a/src/components/AIChat/AIChat.tsx b/src/components/AIChat/AIChat.tsx
index bb06812a5..0fda90bab 100644
--- a/src/components/AIChat/AIChat.tsx
+++ b/src/components/AIChat/AIChat.tsx
@@ -229,15 +229,14 @@ export function AIChat(props: AIChatProps) {
       }
     }
 
-    const reader = response.body?.getReader();
-
-    if (!reader) {
+    const stream = response.body;
+    if (!stream) {
       setIsStreamingMessage(false);
       toast.error('Something went wrong');
       return;
     }
 
-    await readChatStream(reader, {
+    await readChatStream(stream, {
       onMessage: async (content) => {
         const jsx = await renderMessage(content, aiChatRenderer, {
           isLoading: true,
diff --git a/src/hooks/use-roadmap-ai-chat.tsx b/src/hooks/use-roadmap-ai-chat.tsx
index dda428bcd..6f6abbcde 100644
--- a/src/hooks/use-roadmap-ai-chat.tsx
+++ b/src/hooks/use-roadmap-ai-chat.tsx
@@ -188,14 +188,14 @@ export function useRoadmapAIChat(options: Options) {
         return;
       }
 
-      const reader = response.body?.getReader();
-      if (!reader) {
+      const stream = response.body;
+      if (!stream) {
         setIsStreamingMessage(false);
         toast.error('Something went wrong');
         return;
       }
 
-      await readChatStream(reader, {
+      await readChatStream(stream, {
         onMessage: async (content) => {
           if (abortController?.signal.aborted) {
             return;
diff --git a/src/lib/chat.ts b/src/lib/chat.ts
index 4593c8ffc..b1004b865 100644
--- a/src/lib/chat.ts
+++ b/src/lib/chat.ts
@@ -1,10 +1,25 @@
 export const CHAT_RESPONSE_PREFIX = {
-  message: '0:',
-  details: 'd:',
+  message: '0',
+  details: 'd',
 } as const;
 
+const NEWLINE = '\n'.charCodeAt(0);
+
+function concatChunks(chunks: Uint8Array[], totalLength: number) {
+  const concatenatedChunks = new Uint8Array(totalLength);
+
+  let offset = 0;
+  for (const chunk of chunks) {
+    concatenatedChunks.set(chunk, offset);
+    offset += chunk.length;
+  }
+  chunks.length = 0;
+
+  return concatenatedChunks;
+}
+
 export async function readChatStream(
-  reader: ReadableStreamDefaultReader<Uint8Array>,
+  stream: ReadableStream<Uint8Array>,
   {
     onMessage,
     onMessageEnd,
@@ -15,39 +30,62 @@ export async function readChatStream(
     onDetails?: (details: string) => Promise<void> | void;
   },
 ) {
+  const reader = stream.getReader();
   const decoder = new TextDecoder('utf-8');
+  const chunks: Uint8Array[] = [];
+
+  let totalLength = 0;
   let result = '';
 
   while (true) {
-    const { value, done } = await reader.read();
-    if (done) {
+    const { value } = await reader.read();
+    if (value) {
+      chunks.push(value);
+      totalLength += value.length;
+      if (value[value.length - 1] !== NEWLINE) {
+        // if the last character is not a new line, we need to wait for the next chunk
+        continue;
+      }
+    }
+
+    if (chunks.length === 0) {
+      // end of stream
       break;
     }
 
-    const textWithNewLine = decoder.decode(value);
-    const text = textWithNewLine.replace(/\n$/, '');
+    const concatenatedChunks = concatChunks(chunks, totalLength);
+    totalLength = 0;
 
-    if (text.startsWith(CHAT_RESPONSE_PREFIX.message)) {
-      const textWithoutPrefix = text.replace(CHAT_RESPONSE_PREFIX.message, '');
-
-      // basically we need to split the text by new line
-      // and send it to the onMessage callback
-      // so that we don't have broken tags for our rendering
-      let start = 0;
-      for (let i = 0; i < textWithoutPrefix.length; i++) {
-        if (textWithoutPrefix[i] === '\n') {
-          result += textWithoutPrefix.slice(start, i + 1);
-          await onMessage?.(result);
-          start = i + 1;
+    const streamParts = decoder
+      .decode(concatenatedChunks, { stream: true })
+      .split('\n')
+      .filter((line) => line !== '')
+      .map((line) => {
+        const separatorIndex = line.indexOf(':');
+        if (separatorIndex === -1) {
+          throw new Error('Invalid line: ' + line + '. No separator found.');
         }
-      }
 
-      if (start < textWithoutPrefix.length) {
-        result += textWithoutPrefix.slice(start);
+        const prefix = line.slice(0, separatorIndex);
+        const content = line.slice(separatorIndex + 1);
+
+        switch (prefix) {
+          case CHAT_RESPONSE_PREFIX.message:
+            return { type: 'message', content: JSON.parse(content) };
+          case CHAT_RESPONSE_PREFIX.details:
+            return { type: 'details', content };
+          default:
+            throw new Error('Invalid prefix: ' + prefix);
+        }
+      });
+
+    for (const part of streamParts) {
+      if (part.type === 'message') {
+        result += part.content;
+        await onMessage?.(result);
+      } else if (part.type === 'details') {
+        await onDetails?.(part.content);
       }
-    } else if (text.startsWith(CHAT_RESPONSE_PREFIX.details)) {
-      const textWithoutPrefix = text.replace(CHAT_RESPONSE_PREFIX.details, '');
-      await onDetails?.(textWithoutPrefix);
     }
   }