Configuring the AI.

Out of the box the AI sounds like a competent Australian receptionist who knows ServiceM8. These are the four places you shape that into your business: the persona, your vocabulary, your service area, and your knowledge base.

Where to find these settings

Dashboard → Persona for prompt, voice, and vocabulary. Dashboard → Service area for coverage. Dashboard → Knowledge base for work types and FAQs. Changes are live within seconds — no redeploy.

1. Persona prompt

The persona prompt sets the AI's tone and the rules it follows in every conversation. You start from a preset (Classic AU receptionist / Warm and casual / Professional and brief / No-nonsense) and edit from there.

What you can change:tone of voice, the AI's name, opening style, sign-off phrasing, what kinds of questions it should defer to a human instead of answering, business-specific behaviour ("always mention the after-hours surcharge when asked").

What you can't change: the disclosure opener (AI identity + recording line), the call-recording consent, or the DNCR check on outbound calls. Those are compliance invariants — they ship with every tenant.

Versioning. Every save creates a new prompt_version_id. Calls log which version they ran. If you change something and quality regresses, the dashboard shows you exactly which version of the prompt was in play, and you can roll back.

2. Tenant vocabulary

Vocabulary is the layer between speech-to-text and the persona. It catches the words your business uses that a generic model wouldn't.

Preferred phrases

Words you want the AI to use. Useful for terms-of-art your customers expect: "callout fee" not "call-out charge"; "hot water unit" not "water heater".

Banned phrases

Words the AI must not say. Common bans we see: "definitely", "absolutely", "great question", robotic "Goodbye". We ship a few opinionated bans by default (the words that consistently make AI-ish calls sound AI-ish — including a hard ban on "sweet" as filler); you can add more.

Pronunciations

For the speech engine. Suburb names, surnames, and brand names the model mispronounces. "Indooroopilly" → in-droo-ROO-pilly. "Smeg" → smeg(not "S-M-E-G").

How to grow it:when reviewing a call, transcript words you wouldn't use → click Add to banned. Mispronunciations → click Add pronunciation. The vocabulary grows from real calls, not from imagination.

3. Service area

The suburbs or postcodes you cover. The AI uses this to answer "are you in our area?"and — more importantly — to refuse to book work outside coverage instead of silently scheduling a job you can't do.

When a caller's address resolves to a postcode outside your area, the AI reads back the service-area note you've configured (default: "Unfortunately we don't cover that area, but I can take a message in case that changes") and offers a message capture instead of a booking.

Edit your suburbs/postcodes any time. Configure the read-back wording per-tenant so it sounds like you.

4. Knowledge base

The knowledge base is the structured data the AI uses to talk about what you do. It sits in your tenant config as a list of work types, each describing one thing you handle.

What's in a work type

name: Customer-facing label. E.g. "blocked drain".
intent_keywords: Words or phrases customers actually use for this. Drives routing into the right workflow.
capture_fields: What the AI must ask before booking. E.g. for blocked drain: which_drain, since_when, backing_up.
emergency: If true, the AI offers the on-call callout after-hours. Otherwise it takes a message.
opener_phrase: Natural noun phrase used when the AI calls a customer back via SMS thread. E.g. "your blocked drain".
typical_duration_min: What the AI tells the customer to expect for the visit length. Drives the booking window written to ServiceM8.

What it's for

When a customer rings about something on the list, the AI recognises the intent, asks the capture fields in order, applies the emergency rule for after-hours, and books with the right duration. If they ring about something noton the list, the AI captures a message and transfers to a human — it doesn't improvise scope it wasn't briefed on.

Defaults.New tenants ship with a residential trades knowledge base covering 15 common services across plumbing, electrical, HVAC, locksmithing, glazing, pest, gas fitting, drainage, appliance repair, and handyman work. Edit, delete, or add work types — it's your data.

Composition order (for the curious)

When the AI runs, your settings are composed into a single system prompt in a fixed order:

Layer	What it contributes
1. Base	Compliance, banned phrases, opener structure — the invariants that ship with every tenant.
2. Tenant persona	Your persona prompt, AI name, voice selection.
3. Tenant vocabulary	Preferred phrases, banned phrases, pronunciations.
4. Knowledge base	Work types, service area, and the read-back wording for out-of-area calls.
5. Customer context	Pre-warmed from ServiceM8 on phone-number match: last few jobs, open quotes, open invoices.
6. Workflow	The goal, available tools, and prompt for whichever workflow the router picked.

You don't edit this order — it's structural — but knowing it helps when debugging: "why did the AI say X?" traces to one of these six layers, and the conversation review page tags the prompt version it ran.

Common configuration mistakes

Listing every possible service.The knowledge base is what the AI books. Things outside it still get captured (as a message); they just don't get auto-booked. Start narrow.
Banning words you don't actually want banned. Bans cascade — the AI will pick more stilted alternatives. Only ban things that genuinely jar when you hear them.
Forgetting capture fields.If the AI doesn't ask about urgency for a blocked drain, your team has to ring back. Add the capture fields you'd ask if you picked up yourself.
Setting service area too tight. If you sometimes do jobs one suburb over, include it. Out-of-area calls get bounced cleanly; in-area-but-rare calls book.

Workflows — how each conversation type uses your config.
Approval mode & graduation — review prompt changes before they go live for everyone.
Onboarding guide — the first-time setup walkthrough.