An AI front desk that answers — and knows
when to hand off to a human
Concierge embeds into any website with one script tag: routine questions get instant answers from your approved knowledge base; anything uncertain or case-specific escalates to a human — staff reply in Telegram and it lands in the visitor's chat window instantly; every exchange is logged and auditable. AI for speed, humans for accountability — built for law, medical, insurance, and accounting, where a wrong answer carries liability.
- Answers only from your approved knowledge base — never invents facts
- Escalates when unsure — staff reply in a chat, it reaches the visitor instantly
- Every conversation logged, fully auditable
One script tag. That's it.
Copy the tag below and paste it anywhere on your site — Concierge appears in the corner instantly. Shadow DOM isolation means no style clashes. Brand name, colors, AI persona, and knowledge base are all configured in the admin panel, no front-end code required.
<script src="https://chat.nextagent.ca/widget.js" defer data-locale="en"></script>Plug and play — this is exactly how the chat bubble on this site was embedded. This page is the live proof.
Three steps from zero to live
You don't write the FAQ from scratch — AI drafts it, you just review.
Give us your URL
AI crawls your site and drafts a first version of your knowledge base in minutes — services, FAQs, pricing, all covered.
Review and approve in the admin panel
Confirm, edit, or add to the knowledge base. Only approved content is ever used to answer visitors — the AI never improvises.
Go live with one script tag
Paste the tag into your site and Concierge starts working in the corner, 24/7.
Built for industries where a wrong answer carries liability
Not a smarter chatbot — a bounded, auditable customer-service process.
Answers only the approved base — no invented facts
Anything outside the knowledge base isn't guessed at — the AI says "I'm not sure" and escalates instead of making a commitment on your behalf.
Human takeover in Telegram
Escalated conversations push instantly to your Telegram group. Staff reply right there and it lands in the visitor's chat window — no extra admin login, works from a phone.
Lead-capture form
A lead form appears alongside the escalation, capturing name, contact info, and intent in one step — no serious visitor slips through.
Full audit trail
Every message and every model call is logged. Customers get their own admin login to review and audit the full history anytime.
Bilingual EN/ZH
Visitors get answers in Chinese or English automatically, from the same knowledge base — out of the box.
Brand and persona customization
Name, avatar, theme color, and AI tone of voice are all configurable in the admin panel — it feels like your brand, not a generic third-party widget.
Private deployment option
The standard tier is hosted on our infrastructure; for sensitive data or compliance needs, the whole system can run on your own servers.
AI-generated knowledge base
No need to write the FAQ from scratch before launch — AI crawls your site and drafts the knowledge base; you just review and refine it.
Concierge vs. a generic chatbot
Generic bots answer everything to look smart. Concierge answers only what you've approved — and hands the rest to people.
Source of answers
Model improvises; may invent facts
When unsure
Answers anyway; errors surface later
Compliance & audit
No transcript; nothing to review after an incident
Embedding & brand
Third-party page, clashing styles
Source of answers
Only your approved knowledge base; no invented facts
When unsure
Escalates to a human; a Telegram reply reaches the visitor instantly
Compliance & audit
Every message and model call logged and auditable
Embedding & brand
One script tag; brand colors, persona, and both languages configurable
Two ways to deploy — pick what fits
There's no one-size-fits-all price — hosted or private deployment depends on your industry, scale, and compliance needs.
Hosted
Ready to go, hosted on our infrastructure
- One script tag, live in minutes
- AI drafts the knowledge base; you just review and approve
- Telegram human takeover + lead-capture form
- Full audit trail with your own admin login
- Bilingual EN/ZH, brand and persona customization
Private deployment
Runs on your own servers or intranet
- Data never leaves your infrastructure
- Meets compliance needs for medical, legal, and similar industries
- Same knowledge-base review + human-takeover workflow
- Custom integrations for your environment (SSO, internal chat tools, etc.)
Pricing is scoped to your traffic, knowledge base size, and integration needs — book a 15-minute call and we'll give you a concrete range.
Concierge FAQ
Will the AI make things up or give wrong commitments?
No. Concierge answers only from the knowledge base you approve in the admin panel. Anything out of scope or case-specific escalates to a human, with instant Telegram notifications to staff. Every conversation is logged for review — that audit trail is exactly what separates it from a generic chatbot.
How long does setup take? Do I need to change my website?
Add one <script> tag — no changes to your existing code, and Shadow DOM isolation means no style clashes. On first setup, AI crawls your site to draft a knowledge base; you review and approve it to go live. Brand name, colors, and AI persona are all managed in the admin panel, and bilingual EN/ZH works out of the box.
How exactly does human takeover work?
The moment a conversation escalates, staff get notified in a dedicated Telegram group. They reply right there, and that reply appears instantly in the visitor's chat window — no extra admin login required, and it works from a phone.
Who maintains the knowledge base?
On first setup, AI crawls your site and drafts a knowledge base for you — you review, edit, and approve it. After that, you can add or update content anytime in the admin panel as your team or policies change.
Where is data stored? Can it be privately deployed?
The standard tier is hosted on our servers, with conversations and the knowledge base encrypted at rest. If your industry requires data to stay on-premise — medical, legal, and similar fields — the entire system can be deployed privately on your own servers.