Email-Generating AI Integration

I’m enhancing an existing Node.js (Express) back-end with a Vue 3 front-end and want to embed an AI writing assistant focused on producing polished emails and short messages from user prompts.

Here’s how I picture the workflow: a user types a brief prompt (“Follow up about Thursday’s demo”) and selects tone or length options; the system calls an AI service (OpenAI, Claude, or a comparably powerful model), receives the generated draft, and returns it to the Vue component for immediate editing or copy-pasting.

What I need from you
• A clean Node route or service layer that assembles the prompt, hits the chosen AI API, and streams or returns the response.
• A Vue component (or composable) that hooks into that route, displays a loading state, then renders the draft with basic formatting controls.
• Simple parameters for tone, formality, and word count that I can expand later.
• Environment-based config for keys/secrets, plus concise README notes so I can deploy to staging without guesswork.

Tech context
Backend: Node 18, Express, TypeScript.
Frontend: Vue 3 + Vite, Pinia store.
Auth: JWT already in place—just respect the existing middleware.

Acceptance criteria
1. Given a sample prompt, the system returns an email or message within 5 s.
2. Edge cases (empty prompt, API error) surface readable error messages.
3. Code passes eslint & unit tests (vitest/jest) I’ll run on PR.

If you’ve integrated GPT-4, Claude, or similar into JS stacks before, this should be quick; I’m open to your suggestions on prompt engineering or caching strategies. Let’s get a writing assistant live!