← BLOG / SECURITY & QUOTA ARCHITECTURE
Enterprise AI Chatbot Security: Pre-LLM Bot Shields, Visitor Quotas & Daily Email Digests
How cryptographic client proofs, composite IP + token + fingerprint visitor caps, and automated daily email digests protect your AI ROI.
Automated floods are dropped before a single model token is spent
Default per-visitor daily limit across IP + Signed Cookie + Fingerprint
Automated morning inbox summary of all chats and captured leads
- Any public AI chat endpoint without pre-LLM cryptographic verification can have its monthly token budget drained in minutes by a trivial curl script.
- Naive IP-only rate limiting blocks legitimate corporate NAT offices while failing against rotating proxies—composite visitor identity solves both.
- Automated daily email digests keep founders and sales teams informed every morning without logging into a dashboard.
Stopping "Denial-of-Wallet" Attacks Before They Reach the LLM
In the AI era, malicious bots don’t need to crash your server to cause damage—they simply flood your public chat endpoint to burn through your monthly message quota in minutes. ChatNOV neutralizes script floods pre-LLM using five zero-cost edge checks: time-bound X-Chatnov-Guard proofs, hidden honeypot fields, strict Origin allowlists, duplicate-payload rejection, and sliding-window burst limiters. Rejected bot requests consume zero monthly turns.
Composite Per-Visitor Daily Quotas (Default 20 Replies/Day)
Genuine buyers typically resolve their questions and convert within 4 to 8 messages. To prevent a single visitor from treating your commercial widget as a free general-purpose AI sandbox, ChatNOV enforces a Studio-configurable 20 replies/day cap per visitor—tracked across composite IP hashes, HMAC-signed HttpOnly cookies, and browser fingerprints so clearing localStorage doesn’t reset the cap. When reached, the widget politely surfaces your 2-stage lead card.
Daily Summaries & Reports Straight in Your Inbox
Busy founders and revenue teams shouldn’t have to remember another dashboard login every day. ChatNOV brings conversation intelligence directly to your email:
- Free Automated Daily Digest (Enabled by Default): Every day, receive a concise summary of visitor conversations, top questions, and captured leads at zero extra turn cost.
- Optional Instant Full-Transcript Emails (+5 Turns): High-ticket sales teams can toggle instant per-conversation transcripts in Studio—delivering the full dialogue to your inbox moments after a visitor finishes chatting (costing 5 message turns per dispatched transcript email).
SSRF Crawler Hardening & Indirect Prompt-Injection Sandboxing
On the backend, ChatNOV’s document crawler blocks private IP ranges and cloud metadata endpoints (`169.254.169.254`, `127.0.0.0/8`, `10.0.0.0/8`) against SSRF probing, while wrapping all retrieved chunks inside untrusted `
Architectural Comparison Matrix
| Security & Reporting Layer | ChatNOV Guard Architecture (#1) | Unprotected AI Widgets |
|---|---|---|
| Bot & Headless Defense | ✓ X-Chatnov-Guard proof + Honeypot + UA verification | Unprotected public POST endpoint |
| Burst & Duplicate Spam Filter | ✓ Sliding-window burst cap + duplicate payload rejection | Bills full model tokens on spam |
| Visitor Quota Identity | ✓ IP Hash + HMAC HttpOnly Cookie + Browser Fingerprint | Client-side localStorage only (easily bypassed) |
| Inbox Reporting | ✓ Free daily summary digest + Optional instant transcript (+5 turns) | Manual dashboard log checking only |
Frequently Asked Questions
Can we customize the 20 replies/day per-visitor quota?
Yes. Adjust the daily per-visitor limit to any threshold anytime inside ChatNOV Studio.
Do daily summary digest emails consume monthly message turns?
No. Automated daily summary digests are 100% free on every plan. Only optional instant per-conversation full-transcript emails consume 5 turns per email.