Tradie Front Office AI
Tradie Front Office
Northside Plumbing & Electrical
Live business front desk

Omnichannel agent demo

Run one inbound channel, watch it move through the same Kimi K2.6 core, then inspect the normalized proof that lands in the inbox audit trail.

Call demo number Text demo number Inbox audit
Agent runtime
Checking
Checking Workers AI + D1...
Kimi K2.6 + tools
Agent Memory
Checking
D1 docs + chunks
Policy retrieval across channels
Phone and SMS
Checking
+1 (807) 805-3625
Twilio webhooks
Inbound email
Checking
jobs@tradie-front-office-ai.pages.dev
Cloudflare Email worker
Step 1

Choose an inbound channel

Ready

Each run posts to the real route, activates the live flow beside it, then writes proof into the audit panel when D1 is available.

Voice transcript

Twilio speech result into /api/voice/twilio.

voice

SMS message

Twilio-style message into /api/sms.

sms

Email inquiry

Cloudflare Email style payload into /api/email/inbound.

email

Web chat

Dashboard chat payload into /api/chat.

web
Step 2 and 3

Watch the route resolve

Ready
Inbound route
Selected channel enters one intake path.
Agent core
Memory, policy, and task intent are resolved.
Tools and D1
Availability, pricing, job, or escalation runs.
Proof panel
Response, tools, and inbox link appear below.
Inbound
Voice call
Twilio speech transcript
/api/voice/twilio
SMS
Twilio message webhook
/api/sms
Email
Cloudflare Email Worker
/api/email/inbound
Web chat
Dashboard chat widget
/api/chat
Agent core

Kimi K2.6

Guardrailed business concierge. It can answer, ask follow-ups, check D1, quote deterministic prices, book, or escalate.

Model
Workers AI
Memory
D1 RAG
Pricing
Tool-only
Scope
business_id
Outbound proof
D1 inquiry
Conversation and caller details persisted
Agent Memory
Enabled policy chunks retrieved when relevant
Availability
Service window search by job type, suburb, and urgency
Rate card quote
calculatePrice via quote_price only
Job / hold
D1 job with payment pending
Staff escalation
Human handoff when policy requires it
Replay lab

Stress-test the same agent loop

Single scenario cards replay into the cockpit above. Replay All 10 keeps its audit output in the batch summary below.