waitUntil and parallelize (#10)

package.json

@@ -28,6 +28,7 @@
 		"@tailwindcss/vite": "^4.0.0",
 		"@testing-library/jest-dom": "^6.6.3",
 		"@testing-library/svelte": "^5.2.4",
+		"@vercel/functions": "^2.2.0",
 		"clsx": "^2.1.1",
 		"concurrently": "^9.1.2",
 		"convex": "^1.24.8",

pnpm-lock.yaml

@@ -63,6 +63,9 @@ importers:
       '@testing-library/svelte':
         specifier: ^5.2.4
         version: 5.2.8(svelte@5.34.1)(vite@6.3.5(@types/node@24.0.1)(jiti@2.4.2)(lightningcss@1.30.1))(vitest@3.2.3(@types/node@24.0.1)(jiti@2.4.2)(jsdom@26.1.0)(lightningcss@1.30.1))
+      '@vercel/functions':
+        specifier: ^2.2.0
+        version: 2.2.0
       clsx:
         specifier: ^2.1.1
         version: 2.1.1
@@ -1026,6 +1029,15 @@ packages:
     resolution: {integrity: sha512-qHV7pW7E85A0x6qyrFn+O+q1k1p3tQCsqIZ1KZ5ESLXY57aTvUd3/a4rdPTeXisvhXn2VQG0VSKUqs8KHF2zcA==}
     engines: {node: ^18.18.0 || ^20.9.0 || >=21.1.0}
 
+  '@vercel/functions@2.2.0':
+    resolution: {integrity: sha512-x1Zrc2jOclTSB9+Ic/XNMDinO0SG4ZS5YeV2Xz1m/tuJOM7QtPVU3Epw2czBao0dukefmC8HCNpyUL8ZchJ/Tg==}
+    engines: {node: '>= 18'}
+    peerDependencies:
+      '@aws-sdk/credential-provider-web-identity': '*'
+    peerDependenciesMeta:
+      '@aws-sdk/credential-provider-web-identity':
+        optional: true
+
   '@vitest/expect@3.2.3':
     resolution: {integrity: sha512-W2RH2TPWVHA1o7UmaFKISPvdicFJH+mjykctJFoAkUw+SPTJTGjUNdKscFBrqM7IPnCVu6zihtKYa7TkZS1dkQ==}
 
@@ -3224,6 +3236,8 @@ snapshots:
       '@typescript-eslint/types': 8.34.0
       eslint-visitor-keys: 4.2.1
 
+  '@vercel/functions@2.2.0': {}
+
   '@vitest/expect@3.2.3':
     dependencies:
       '@types/chai': 5.2.2

src/routes/account/models/model-card.svelte

@@ -39,7 +39,6 @@
 
 	async function toggleEnabled(v: boolean) {
 		enabled = v; // Optimistic!
-		console.log('hi');
 		if (!session.current?.user.id) return;
 
 		const res = await ResultAsync.fromPromise(

src/routes/api/generate-message/+server.ts

@@ -7,6 +7,7 @@ import { error, json, type RequestHandler } from '@sveltejs/kit';
 import { ConvexHttpClient } from 'convex/browser';
 import { ResultAsync } from 'neverthrow';
 import OpenAI from 'openai';
+import { waitUntil } from '@vercel/functions';
 
 import { z } from 'zod/v4';
 
@@ -48,14 +49,30 @@ async function generateAIResponse(
 ) {
 	log('Starting AI response generation in background', startTime);
 
-	const modelResult = await ResultAsync.fromPromise(
+	const [modelResult, keyResult, messagesQueryResult] = await Promise.all([
-		client.query(api.user_enabled_models.get, {
+		ResultAsync.fromPromise(
-			provider: Provider.OpenRouter,
+			client.query(api.user_enabled_models.get, {
-			model_id: modelId,
+				provider: Provider.OpenRouter,
-			session_token: session.token,
+				model_id: modelId,
-		}),
+				session_token: session.token,
-		(e) => `Failed to get model: ${e}`
+			}),
-	);
+			(e) => `Failed to get model: ${e}`
+		),
+		ResultAsync.fromPromise(
+			client.query(api.user_keys.get, {
+				provider: Provider.OpenRouter,
+				session_token: session.token,
+			}),
+			(e) => `Failed to get API key: ${e}`
+		),
+		ResultAsync.fromPromise(
+			client.query(api.messages.getAllFromConversation, {
+				conversation_id: conversationId as Id<'conversations'>,
+				session_token: session.token,
+			}),
+			(e) => `Failed to get messages: ${e}`
+		),
+	]);
 
 	if (modelResult.isErr()) {
 		log(`Background model query failed: ${modelResult.error}`, startTime);
@@ -70,13 +87,7 @@ async function generateAIResponse(
 
 	log('Background: Model found and enabled', startTime);
 
-	const messagesQuery = await ResultAsync.fromPromise(
+	const messagesQuery = await messagesQueryResult;
-		client.query(api.messages.getAllFromConversation, {
-			conversation_id: conversationId as Id<'conversations'>,
-			session_token: session.token,
-		}),
-		(e) => `Failed to get messages: ${e}`
-	);
 
 	if (messagesQuery.isErr()) {
 		log(`Background messages query failed: ${messagesQuery.error}`, startTime);
@@ -86,14 +97,6 @@ async function generateAIResponse(
 	const messages = messagesQuery.value;
 	log(`Background: Retrieved ${messages.length} messages from conversation`, startTime);
 
-	const keyResult = await ResultAsync.fromPromise(
-		client.query(api.user_keys.get, {
-			provider: Provider.OpenRouter,
-			session_token: session.token,
-		}),
-		(e) => `Failed to get API key: ${e}`
-	);
-
 	if (keyResult.isErr()) {
 		log(`Background API key query failed: ${keyResult.error}`, startTime);
 		return;
@@ -271,9 +274,11 @@ export const POST: RequestHandler = async ({ request }) => {
 	}
 
 	// Start AI response generation in background - don't await
-	generateAIResponse(conversationId, session, args.model_id, startTime).catch((error) => {
+	waitUntil(
-		log(`Background AI response generation error: ${error}`, startTime);
+		generateAIResponse(conversationId, session, args.model_id, startTime).catch((error) => {
-	});
+			log(`Background AI response generation error: ${error}`, startTime);
+		})
+	);
 
 	log('Response sent, AI generation started in background', startTime);
 	return response({ ok: true, conversation_id: conversationId });