Add whoopsie to your AI app.

Add "whoopsie" failure observability to my AI app.

What it does: catches when my chatbot loops, hallucinates, or burns through tokens, and shows the failures live at https://whoopsie.dev/live/ws_b9oPiKAACu4GwjKY

Steps to follow exactly:

1. Add the dependency: `@whoopsie/sdk`.

2. Find the file in my project that calls `streamText` or `generateText` from the `ai` package. It's usually `app/api/chat/route.ts`. If there are multiple, pick the main chat one.

3. At the top of that file, make sure these two imports exist (add the second one):

   ```ts
   import { wrapLanguageModel } from "ai";
   import { whoopsieMiddleware } from "@whoopsie/sdk";
   ```

4. Find the line that passes a model. It looks something like `model: openai("gpt-4o")` or `model: anthropic("claude-sonnet-4-6")`. Wrap the existing model expression with `wrapLanguageModel`:

   ```ts
   model: wrapLanguageModel({
     model: <whatever was there before>,
     middleware: whoopsieMiddleware(),
   })
   ```

5. Add this environment variable to Lovable → Project Settings → Environment Variables:

   ```
   WHOOPSIE_PROJECT_ID=ws_b9oPiKAACu4GwjKY
   ```

6. After you've made the changes, reply to me with one sentence:
   "Whoopsie is wired up. Open your live dashboard at https://whoopsie.dev/live/ws_b9oPiKAACu4GwjKY and hit your chat once."

Do not skip step 5 (the env var) or step 6 (the confirmation message).