What's actually happening here
Emily comes from the world of workflows and automation — and she built emAI the way she'd build anything else: as a real system, not a bolted-on widget. It's a small end-to-end pipeline working hand-in-hand with AI tools. Here's the flow, from your browser to the AI and back — told the way she'd actually track it: as a ticket.
Ticket lifecycle
Submitted
You type a question or tap a suggested prompt. It leaves your browser as a secure request, locked down so only this site can call the backend.
Your browser, HTTPS + CORS-restrictedRouted
An edge function checks the request, enforces a per-visitor rate limit, and attaches Emily's background as context. The Anthropic API credential lives only here — it never touches your browser.
Cloudflare WorkerResolved
Claude writes a reply using only the attached background material — not whatever it might already know or guess about Emily Pham from elsewhere.
Claude, single source of truthClosed
The reply streams back word by word and lands in the chat — ready for whatever you ask next.
Service-level commitments
Stays in scope
If a question falls outside the scope of Emily's candidacy, the bot declines to guess and instead invites the user to continue the conversation with her directly on LinkedIn.
Capped on purpose
A per-minute message limit and a length cap per reply keep the assistant available to everyone and the API cost predictable.
One document, no embellishment
Every fact traces back to a single reviewed background document. Nothing about Emily's experience is invented, inflated, or pulled in from elsewhere.