This is what a customer sees at 11pm on a Sunday.
It knows your services, your hours, and your prices, and answers before your customer gives up and calls a competitor instead. And when an off-the-shelf widget isn't enough, our team builds bespoke engagements too: WhatsApp bots, voice receptionists, document extraction.
A fallback model is always waiting.
Our AI chatbot runs three layers deep: a primary model answers most queries, a fallback model activates if it times out, and a deterministic safety net catches the rest. See the full architecture on /platform →
The same layered approach runs everywhere AI touches the platform, including document and data extraction (see below): a cost-efficient model handles the routine case, a stronger multimodal model steps in when needed, and a deterministic fallback catches whatever's left.
How a turn gets routed
-
On-network inference Routine turns
Everyday questions are answered by models running on the same edge network as your site, milliseconds from your visitor, at about $0.01 a conversation.
-
Frontier reasoning Hard questions
Anything genuinely complex escalates to a frontier reasoning model, so the hard 5% of conversations are handled properly instead of cheaply.
-
Multimodal understanding Beyond text
Documents, photos and screenshots are read as readily as text, which is what makes intake forms, uploaded records and photo-led workflows possible.
-
Automatic fallback Reliability
Every AI feature has a second provider already holding the conversation and a scripted response behind that, so a provider outage degrades the answer instead of removing the feature.
The right tier is picked per task, priced at what our provider charges us. A second provider keeps things running even during an outage.
You see the exact AI cost the same day it happens.
Every AI feature we build tracks its exact cost per use, with a spending ceiling you control.
Metered in real time
Every AI call's real token usage and cost is recorded the moment it happens, the number you see is exactly what occurred.
A ceiling you set
Cap spend on the higher-cost fallback model if you want, routine work keeps running on the efficient model, nothing silently escalates past your limit.
Edge-cheap by default
Routine AI work runs on the same edge network as your site at $0.01 a conversation, milliseconds from your visitor's location.
# robots.txt (excerpt)
User-agent: GPTBot
Allow: /
User-agent: ClaudeBot
Allow: /
User-agent: PerplexityBot
Allow: /
# ...and more, verified allow-list, not a wildcard.
Show up when ChatGPT, Perplexity and the rest get asked.
Many platforms now block the AI assistants' crawlers by default, without anyone realizing it. Every site we build ships with an explicit, named AI-crawler allow-list, so verified assistants can read it while scraping bots stay blocked, plus an AI-readable summary of the business kept in sync with the live site.
We hold ourselves to the same standard: this site's own AI-readable summary is public at /llms.txt, and this exact package, allow-list, structured data, real-time indexing, is what ships in AI Search & Answer Engine Visibility.
Every engagement is priced the same three-part way.
Every AI engagement, chatbot, WhatsApp bot, voice receptionist, or something bespoke - is priced the same way: fixed setup + usage at provider cost + Managed Platform fee. You know all three before any work begins, usage carries no markup, and the fee covers the operations on top: monitoring, audit logs, AI rate limiting and incident response.
Scope the work
Tell us what you need on a short call. We map it to your stack and send a fixed setup quote within 48 hours.
Build and approve
Our team designs, builds and tests the AI engagement on your platform, with demos along the way. You approve before anything goes live.
We operate it
Monitoring, AI rate limiting, audit logs, updates and incident response are covered by the Managed Platform fee, with usage priced at what our provider charges us.
The four builds businesses ask us for most.
AI Chatbot (Web Widget)
Your 24/7 front desk, answers, qualifies & books.
- Trained on your services, prices & FAQs automatically from your CMS
- Typically responds in seconds, around the clock
- Knowledge base managed directly from your CMS
- Auto-escalates to a human staff member when needed
- Can collect booking enquiries & contact details
- Runs on our managed AI runtime, with a fallback model for reliability
- Full conversation history in your dashboard
- Works on mobile and desktop without any app
WhatsApp AI Bot
The same AI on your own WhatsApp number, built to your needs.
- Runs on your own verified WhatsApp Business number
- Inbound replies within the 24-hour service window, no per-message platform fee
- Shares the same knowledge base as the web chatbot
- Human handoff with full conversation context
- Message history synced to your operations dashboard
- Outbound template messages billed by the messaging provider (passed through at cost)
AI Voice Receptionist
An AI that answers your calls, qualifies and books, 24/7.
- Answers calls in a natural voice, around the clock
- Shares the same knowledge base as your chatbot
- Captures bookings, enquiries & callbacks
- Warm handoff to a human when needed
- Call transcripts & summaries in your dashboard
- Per-minute telephony billed at cost
AI Document & Data Extraction
Business cards, forms, contracts, IDs and receipts, pulled into structured data automatically.
- Pulls structured data from documents you upload, no manual retyping
- Cost-efficient model handles extraction first; a stronger multimodal model steps in only when needed
- Every extraction logged with its real cost, same metered accounting as the chatbot
- A human confirms the result before it's used
Need something beyond AI, custom logins, compliance, integrations, or scaling? See every tailored engagement on /pricing →
A live one is answering customers right now.
Bilingual, trained on the business's real services and prices, answering around the clock and escalating to staff when it matters. Open it, ask it something hard, and imagine it trained on your business instead.
Try the live chatbotWhere AI shows up across the platform.
AI Chatbot
Your 24/7 front-line assistant, trained on your business, escalates to your team when it matters.
Learn more →AI Content Assistant
Drafts service copy, FAQs and translations in your voice, you approve before anything publishes.
Learn more →AI Smart Search
Visitors find the right service by meaning, synonyms, typos and vague descriptions all resolve correctly.
Learn more →AI Search & Answer Engine Visibility
Built to be read and cited by the major AI assistants and answer engines, the same way it's built to rank on Google.
Learn more →Bespoke AI engagements
WhatsApp bots, voice receptionists, document extraction, built for your business specifically.
Learn more →AI questions
How is AI work routed, and why more than one model?
What happens if an AI provider has an outage?
Can you build a custom AI tool for my business beyond a chatbot?
Does adding AI make my site slower, or unpredictable to budget for?
Why aren't AI features included in the plans?
What login and account systems can you set up?
What does an AI engagement cost?
Book a 30-minute discovery call.
We look at your current setup, walk you through the live platform, and send a written quote within 48 hours. No obligations.
We reply within one business day. Available in English and Spanish.