First AI job
AI Call Concierge
Answer, qualify, book, summarise — every call, day or night.
For Pembroke Plumbing, the answer is not “which AI tool” — it is giving AI one job worth money: answer every call, take the details, book the job, and flag emergencies the moment they come in.
Dawn's problem isn't lack of tools. It's that calls slip when her hands are full — and slipped calls become someone else's jobs. The right first AI job sits where money enters the business, the work repeats every day, and a human can review what the AI offered before it commits.
Sam's job is to turn a busy owner's AI overwhelm into one usable operating decision. The conversation is deliberately short — owners need the few questions that expose where the leverage actually sits.
Before tools — where does money most often leak after someone's already shown buying intent?
Calls, easily. Phone rings while I'm under a boiler or driving. Half go to voicemail. Most don't leave a message — they just ring the next plumber.
What are the repeatable steps you wish someone handled every time, even when you're busy?
Answer the phone in our name. Take what's broken, when, postcode, urgency. Book it or flag it as an emergency. Just so the call doesn't die.
What should AI never do without your approval?
Quote a price. Promise a slot I haven't checked. Anything contractual.
Good. Your first AI job isn't a “smart assistant” — it's an always-on receptionist. It protects the revenue already trying to reach you.
This is the assessment spine. It keeps the report from becoming a generic AI brain dump and forces the output toward one valuable first job.
Which part of the firm creates or protects revenue?
→ Phone enquiries — the calls she catches become work.
What happens every day, follows a pattern?
→ Taking name, postcode, what's broken, urgency.
What loses value when it waits?
→ A call-back an hour later — already booked elsewhere.
Separate what AI can prepare from what a human must approve.
→ Pricing, time promises, emergencies — all human.
Pick one measurable outcome.
→ Missed calls converted into booked jobs.
The goal isn't every possible AI use case. It's one job with enough revenue, low enough risk, and fast enough proof.
This is the one-sentence answer Dawn should remember. Everything else in the report supports this decision.
The concierge picks up every incoming call, day or night — in the firm's name, in a natural voice. It takes the details (problem, postcode, urgency), books the job into the diary, and texts both the customer and Dawn a clean summary. Emergencies go straight to her mobile.
A narrow job becomes valuable when the handoffs are explicit — what the AI handles, what it escalates, what stays human.
The important design choice: AI is placed BEFORE Dawn picks up, not inside her pricing or scheduling judgement.
Use the sliders to ground this in real numbers. These are starting assumptions to validate in the first week — the strongest case starts when Dawn corrects them.
Dawn doesn't care which model or framework runs it. She cares whether the job gets done, what it costs, and whether emergencies still reach her phone.
Short enough to ship, structured enough to prove the case.
Service area, working hours, FAQ answers (boiler types, emergency criteria), greeting in your voice. Write the escalation rules first — what counts as an emergency, what reaches your mobile.
Live test calls — boiler emergency, kitchen tap, time-waster, out-of-area. Tune the responses, fix the rough edges, check how the texts read.
Concierge takes calls when you don't pick up in four rings. You watch what comes through and how the summaries land. Adjust anything that doesn't sit right.
Concierge answers everything; you decide each evening on flagged jobs. Compare the fortnight against the previous one — calls captured, jobs booked, emergencies routed.
Premium AI work shouldn't pretend the risk is zero. It should show exactly how each risk is contained.
A good assessment leads to one of two clean actions: try it free this week, or have us set it up properly.
Spend a week logging every call you miss — phone, time, did they leave a voicemail. The number we model becomes provable.
DIY Clara/Norango free trial, or have ScaleUP set up and manage it. Decision lives at the bottom of this report.
What counts as an emergency in your firm. What questions you want every caller asked. Where the texts should land.
Off-the-shelf — start the free trial today. Managed — we can have it live within 14 days.
Pembroke Plumbing's first AI move should be an always-on AI Call Concierge — an off-the-shelf service that picks up every call in the firm's name, takes the job details, books the diary slot, and texts Dawn a clean summary. Emergencies route straight to her mobile. Live in days for ~£50/month, with a free trial first. Don't quote pricing or commit to dates without Dawn's approval — those stay human. The measure of success: missed calls converted into booked jobs in the first fortnight.
A useful assessment names one job, one number, one route. Anything else is a brochure.
One named job (Call Concierge), one workflow, one pilot, one primary metric — not a list of AI possibilities.
The ROI model forces real numbers — missed calls, jobs, value and margin — not tool novelty.
AI prepares; humans keep pricing, judgement and any commitment to the customer.