Simulate a real conversation with the user

src/app/actions.ts

@@ -6,10 +6,12 @@ import { createStreamableValue } from 'ai/rsc';
 
 export async function recommend(messages: Message[]) {
 
+  const prompt = messages[messages.length - 1].content
+
   const inference = new HfInference(process.env.HUGGINGFACE_TOKEN);
   const embedding = await inference.featureExtraction({
     model: "sentence-transformers/all-MiniLM-L6-v2",
-    inputs: messages[messages.length - 1].content,
+    inputs: prompt,
   });
 
   // Query the database for similar watches
@@ -38,23 +40,56 @@ export async function recommend(messages: Message[]) {
   const stream = createStreamableValue('');
   (async () => {
 
-    // Update the stream with all the watches[0] properties in a markdown table
-    stream.append(toMarkdown(watches[0]))
-    stream.append(`\n\n\n\n`)
+    const watch = toMarkdown(watches[0])
+    const messages = [
+      {
+        'role': 'user',
+        'content': `You are an helpful AI that will recommend watches to the user. Your responses are formatted as Markdown documents.` + prompt,
+      },
+      {
+        'role': 'assistant',
+        'content': `
+Here is the watch that I would recommend:
 
+${watch}
+        `,
+      },
+      {
+        'role': 'user',
+        'content': `Based on my initial request, can you very briefly justify why you think this watch is a good fit for me in one paragraph?`
+      }
+    ]
+    let answer = ''
     for await (const chunk of inference.chatCompletionStream({
       model: "mistralai/Mistral-7B-Instruct-v0.2",
-      messages: [
-        {
-          'role': 'user',
-          'content': `You are an helpful AI that will recommend watches to the user. The watch selected is below with its attributes as a list. Your response is a markdown document. Write a paragrah of why this watch matches their request using the watch attributes. The recommended watch is: ${toMarkdown(watches[0])}. Create a table of the following data titled as alternatives: ${watches.map((watch: Watch) => extract(watch)).join('\n')}`
-        }
-      ],
+      messages: messages,
       max_tokens: 500,
       temperature: 0,
     })) {
       stream.update(chunk.choices[0].delta.content!);
+      answer += chunk.choices[0].delta.content!;
     }
+
+    messages.push({
+      'role': 'assistant',
+      'content': answer,
+    });
+
+    // Update the stream with all the watches[0] properties in a markdown table
+    stream.update(`\n\n---\n` + watch + `\n---\n`)
+    messages.push({
+      'role': 'user',
+      'content': `Create a table of the following data titled as alternatives: ${watches.map((watch: Watch) => extract(watch)).join('\n')}. Don't add any other information. Do not comment.`,
+    });
+    for await (const chunk of inference.chatCompletionStream({
+      model: "mistralai/Mistral-7B-Instruct-v0.2",
+      messages: messages,
+      max_tokens: 500,
+      temperature: 0,
+    })) {
+      stream.update(chunk.choices[0].delta.content!);
+    }
+
     stream.done();
   })();