Empire Chatbot — Docs

1. Sign up + get embed code

Visit /signup. Get back a single <script> tag.

2. Paste it on your site

Anywhere before </body>. The floating bubble appears bottom-right on every page.

<script src="https://chatbot.eliteagenticsolutions.com/v1/embed/<tenant_key>/widget.js"
        data-token="<embed_token>" defer></script>

3. Upload knowledge sources (Growth+)

From your dashboard, upload Markdown / PDF / URLs. We embed via free HF inference and store in pgvector. The bot answers from your content first; falls back to general knowledge.

4. Compliance modes (Enterprise)

• FERPA: never collects PII from minors; refers academic advice to teachers/counselors.
• CFTC 4.41(b): never provides trading advice; appends hypothetical-performance disclaimer to substantive answers.
• HIPAA-aware: refuses to give medical advice; refers to licensed professionals. (Not a HIPAA Business Associate — never collect PHI.)

5. API (Enterprise)

POST https://chatbot.eliteagenticsolutions.com/v1/embed/<tenant_key>/message
Headers: X-Embed-Token: <embed_token>
Body: {"message": "Hi", "session_id": "user-123"}
Returns: {"reply": "...", "tier": "enterprise", "step": "ok"}

6. Rate limits

Free: 10/day · Starter: 100/day · Growth: 500/day · Enterprise: 2000/day. Counted per visitor-IP.

7. Routing transparency

Every LLM call goes through the F1340 cascade: Groq Fast → Gemini Text → Cerebras → Cloudflare → Gemma → (paid tiers add Cerebras Paid → OpenRouter US-only → OpenAI mini → Together). Anthropic is NEVER called. Chinese-jurisdiction LLMs (DeepSeek, Qwen, Moonshot, etc.) are blocked at the regex layer.