Security

AI with
guardrails.

Every AI response passes through our Reply Guard system. 11 safety rules. Zero hallucinations. Full control over what your AI says.

Customer Message

Incoming question or request

AI Drafts Reply

Generates response using your data

Guard Checks 11 Rules

Every reply scanned in real-time

Pass / Fail

Approved or held for review

Send or Hold

Safe replies sent instantly

✓ Pass -- sent to customer instantly
✗ Fail -- held for human review
11 rules. Every response.
01 No fabricated product info
02 No made-up policies
03 No unauthorized discounts
04 No competitor mentions
05 No personal data exposure
06 No off-topic responses
07 No legal/medical advice
08 No aggressive sales tactics
09 No contradicting store info
10 No unverified claims
11 Tone matches brand voice
Defense in depth.

Reply Guard

11 safety rules checked on every single response. Blocks hallucinations, unauthorized discounts, and off-brand messaging before they reach your customers.

Shadow Mode

High-risk replies are held for human review instead of being sent automatically. You see the AI's draft and approve or edit before the customer sees it.

No Hallucination Policy

The AI never fabricates product information, pricing, or policies. If it doesn't know, it says so and escalates to a human. No guessing, no inventing.

Data Encryption

All data encrypted in transit (TLS 1.3) and at rest (AES-256). Customer conversations, your product data, and business information are fully protected.

SOC 2

We're pursuing SOC 2 Type II certification to meet enterprise security requirements. Audit in progress.

Coming Q3 2026

GDPR Ready

Full GDPR compliance. Data deletion on request, data export, consent management, and EU data residency options. Your customers' privacy is protected.

Learn more about our safety.

Talk to our team →