Most site bots answer from your content. Concierge does too — then adds the part they skip: built-in PII redaction, crisis safeguarding, and a safety layer that can’t be switched off. The trust controls enterprise suites charge per-resolution for — lead-capturing, domain-locked, and live the same day.
Accurate answers, effortless install, and lead capture from day one — the capabilities that turn a visitor into a customer. (Safety & compliance get their own section below.)
The assistant responds strictly from the pages you train it on — no invented facts. It auto-detects your business details and matches your voice.
Capture enquiries and forward them to Zapier, your CRM, or Slack over HMAC-signed webhooks — or by email. Nothing slips through.
Pick exactly which pages to learn from. Retrain only the pages that changed to save tokens, and login or private pages are auto-excluded. Scheduled retraining keeps answers current as your site evolves.
Feed the assistant price lists, policies, and FAQs directly — CSV, Excel, and PDF uploads, or a linked Google Sheet. Uploaded knowledge survives retrains and can be replaced file-by-file.
Context-aware quick-reply buttons keep visitors moving, and a gentle proactive teaser invites them in — a small bubble, never a screen takeover.
A polite “did that answer your question?” feeds your resolution-rate metric. You choose when it appears — mid-conversation or only when the chat is closed.
The assistant defaults to your site’s language and automatically switches to the visitor’s when they write in another one.
One snippet drops into WordPress, Shopify, Wix, Squarespace, or any HTML page — and a one-click WordPress plugin installs it for you.
Paste your URL, choose your pages, set your guardrails, and drop in one line of code. Your assistant is answering questions the same day — and until it’s trained, it shows your holding message instead of guessing.
Your brand, your voice, your rules — and if you want, your own AI model behind it.
Colours, shape, position, light or dark, greeting text, launcher style — every visual is yours to set. Upload bot avatars that even change with the conversation state.
Choose a personality template, tune tone and behaviour per section, and add your own canned responses for the questions you answer the same way every time.
Start from a proven template, then keep saved versions of your configuration — test a variant, roll back in one click. Power-user depth, hidden until you want it.
Run on our managed models, or connect your own LLM endpoint — any OpenAI-compatible API, including self-hosted models. Set a fallback provider chain so an outage never silences your assistant.
Offline message, holding message, error message, greeting, disclosure line — all editable, so the assistant never says anything you wouldn’t.
Refer or manage customers through a dedicated partner portal with revenue attribution — agency and white-label tiers are on the roadmap below.
The protections you’d expect to bolt on later are part of the platform — applied to every conversation, automatically.
Built-in detectors for email, phone, card, and national-ID numbers strip personal data before anything is logged or forwarded. Ready-made compliance packs for UK, EU, and US — general and healthcare — plus your own custom patterns.
If a visitor appears to be in distress, the assistant surfaces emergency contacts — 999, Samaritans, NHS 111 — rather than attempting an answer. This step is locked on: it can’t be disabled, even by an admin.
A global safety layer is applied after every prompt compiles. Toggle off medical, legal, and financial advice — the layer itself can’t be bypassed by any configuration, template, or custom prompt.
Visitors who try to jailbreak, extract the system prompt, or steer the assistant off-script are detected and politely refused — before the AI ever sees the attempt.
Behaviour adapts to your operating regions and industries. Visitors can end a chat and erase their messages in one click (GDPR Articles 15 & 17), and account deletion cascades through every stored record.
Choose how long conversations are kept — 30, 90, or 365 days, or don’t store them at all. An automatic sweep enforces the window; no IP address or visitor identity is ever collected.
The embed runs only on the domains you allow. A copied key on any other site is rejected, so no one can drain your token budget — and per-site message caps put a hard ceiling on spend.
The database accepts connections only from our platform’s fixed IP with least-privilege credentials; outbound webhooks are SSRF-guarded; passwords are hashed with bcrypt and admin accounts support MFA. All traffic is encrypted in transit.
Administrative actions and conversation-data access are logged to an audit trail with its own retention policy — so you can answer “who saw what, when” if a regulator ever asks.
The platform runs on auto-scaling cloud infrastructure designed so a traffic spike is a good day, not an outage.
Runs on Google Cloud with stateless instances that scale out under load and back down when quiet — you never pay us more for a busy week.
Retraining, retention sweeps, and migrations are leader-elected across instances — exactly one runner, no duplicated work, automatic takeover if an instance is recycled.
If a usage cap is reached, the assistant falls back to your FAQs and canned answers instead of going dark — and provider failover chains keep replies flowing through an AI outage.
Updates roll out with health-checked revisions and instant rollback — your assistant keeps answering while we ship.
In active development — delivered to every customer as it lands, at no extra cost.
Connect OneDrive, SharePoint, or Google Drive and pick a folder — update the spreadsheet in your drive and the assistant stays current automatically.
Visitors react to any reply with a tap — instant per-message quality signals without interrupting the conversation.
See your top unanswered questions, resolution trends, and auto-suggested FAQs — the assistant tells you what your website is missing.
Beyond answering: look up orders, build baskets, take bookings, and start refunds through secure connectors to your existing systems.
Manage all your clients’ assistants from one dashboard — and at the white-label tier, under your own brand end to end.
Basic site bots skip the safety layer. Enterprise support suites have it — but charge per resolution and take weeks to onboard. Concierge is the column with checks in every row.
| Capability | Generic site chatbotsRAG widgets, self-serve | Enterprise support suitesper-resolution pricing | Conciergesafety + self-serve |
|---|---|---|---|
| Answers only from your trained content | ✓ | ✓ | ✓ |
| Safety layer on every reply that can’t be bypassed | × | ~ | ✓ |
| Built-in PII redaction before anything is logged | × | ~ | ✓ |
| Crisis & safeguarding detection | × | × | ✓ |
| Toggle off medical / legal / financial advice | × | ~ | ✓ |
| Region-aware GDPR / HIPAA + one-click erasure | ~ | ✓ | ✓ |
| Domain-locked embed key | ~ | ✓ | ✓ |
| Lead capture + email & HMAC-signed webhooks | ~ | ✓ | ✓ |
| Knowledge from files & Google Sheets, not just pages | ~ | ✓ | ✓ |
| Bring your own LLM endpoint (self-hosted / OpenAI-compatible) | × | × | ✓ |
| Live the same day — no sales call | ✓ | × | ✓ |
| Flat pricing — no per-resolution fees | ✓ | × | ✓ |
✓ built in · ~ varies by vendor, an add-on, or enterprise-tier only · × not offered
Flat monthly pricing — never per conversation, per resolution, or per seat. A busy month should be a good month, not a surprise invoice.
A single predictable price. Grow your traffic without watching a meter — you’re never billed by the answer or by the seat.
Enterprise support bots bill roughly £1–2 every time they reply. We don’t — so a viral week can’t blow your budget.
Monthly rolling. Leave whenever you like and take your data with you — we’d rather keep you with the product than with paperwork.
No coding. No complicated setup. Just paste, train, and go.
Create your free account and enter your website URL. Our wizard guides you through everything.
Pick exactly which pages to train on, set your safety toggles and PII rules, and lock the embed to your domain.
Add one line of code to your site. Your AI assistant is live and answering questions instantly.
Tell us about your business and what you’d like your assistant to do. We reply personally, usually within one working day.