🤖 Little Johnny's First AI App

App #1 — Ask anything

The simplest possible AI app. One button. Llama 3 8B answers.

App #2 — Free weather, no LLM tokens

Routes through the realm's weather intent, not the LLM. Free, structured, fast.

App #3 — Today's news, JSON-ready

Hits /api/news. Headlines come back as JSON. Render however you want.

App #4 — Is the realm alive?

The boring-but-essential health check. /api/setup_status.

The source code for App #1 — all of it

<input id="q1" />
<button onclick="ask1()">Ask Llama 3</button>
<div id="out1"></div>
<script>
async function ask1() {
  out1.textContent = "thinking...";
  const r = await fetch("https://rentahal.com/api/ask", {
    method: "POST",
    headers: { "Content-Type": "application/json" },
    body: JSON.stringify({ text: q1.value })
  });
  const data = await r.json();
  out1.textContent = data.answer;
}
</script>

That's the whole app. Save it as index.html, open in any browser, you're done. No npm. No build. No keys.

What this proves

• You can build a real AI app in under 20 lines of HTML
• The backend is a single RTX 4060 in someone's lab
• The LLM is open-source Llama 3 8B, running locally on that GPU
• No cloud middleman, no token costs, no API keys
• Works from any device, any network, any browser
• This page is App #1 itself — view source, learn, fork

Little Johnny built his first AI app. Now it's your turn.