diff --git a/src/graphql/mutations/CreateAIReply.js b/src/graphql/mutations/CreateAIReply.js
index effe0c59..f72d6e8e 100644
--- a/src/graphql/mutations/CreateAIReply.js
+++ b/src/graphql/mutations/CreateAIReply.js
@@ -89,7 +89,7 @@ export async function createNewAIReply({
     ...completionOptions,
   };
 
-  const { update: updateAIResponse } = createAIResponse({
+  const { update: updateAIResponse, getAIResponseId } = createAIResponse({
     user,
     docId: article.id,
     type: 'AI_REPLY',
@@ -97,7 +97,10 @@ export async function createNewAIReply({
   });
 
   // Resolves to completed or errored AI response.
-  const apiResult = await getOpenAI()
+  const apiResult = await getOpenAI({
+    traceId: await getAIResponseId(),
+    traceName: `AI Reply for article ${article.id}`,
+  })
     .chat.completions.create(completionRequest)
     .catch((error) => {
       console.error(error);
diff --git a/src/graphql/util.js b/src/graphql/util.js
index fe359ba5..fdbfde3c 100644
--- a/src/graphql/util.js
+++ b/src/graphql/util.js
@@ -781,7 +781,7 @@ function extractTextFromFullTextAnnotation(fullTextAnnotation) {
 export async function createTranscript(queryInfo, fileUrl, user) {
   if (!user) throw new Error('[createTranscript] user is required');
 
-  const { update } = createAIResponse({
+  const { update, getAIResponseId } = createAIResponse({
     user,
     type: 'TRANSCRIPT',
     docId: queryInfo.id,
@@ -823,7 +823,10 @@ export async function createTranscript(queryInfo, fileUrl, user) {
         // Ref: https://github.com/openai/openai-node/issues/77#issuecomment-1500899486
         const audio = ffmpeg(fileResp.body).noVideo().format('mp3').pipe();
 
-        const data = await getOpenAI().audio.transcriptions.create({
+        const data = await getOpenAI({
+          traceId: await getAIResponseId(),
+          traceName: `Whisper transcript for media ${queryInfo.id}`,
+        }).audio.transcriptions.create({
           // Ref: https://github.com/openai/openai-node/issues/77#issuecomment-2265072410
           file: await toFile(audio, 'file.mp3', { type: 'audio/mp3' }),
           model: 'whisper-1',