Most service calls go unanswered. Here's what happens when you stop missing them.

An after-hours answering service AI is a voice agent that picks up the calls your team misses in the evenings and on weekends. Industry data from ServiceTitan and AInora puts 25–40% of inbound home-service calls outside the 9–5 window — plumbing reaching 30–45% due to emergency leaks. More broadly, 411 Locals (2024, 85 SMBs) found only 38% of small-business calls were answered live, meaning the majority of calls are lost to voicemail or silence. For HVAC, plumbing, and cleaning firms running 5–50 technicians, a single missed call costs $350–1,200 (industry estimates: Invoca 2024 pegs the home-services average at ~$1,200 per missed call; ServiceTitan data supports a $350+ floor for standard HVAC/plumbing jobs). The agent answers, qualifies the job, and routes the urgent ones to a human on call — instead of letting the work walk to whoever picks up first.

A Bank Holiday weekend, in real numbers

Picture a long weekend. Your office is closed Saturday, Sunday, and the Bank Holiday Monday. Over those three days the line rings 23 times and nobody answers. Six of those callers had emergencies — a burst pipe, a dead boiler, a flooded kitchen — and they couldn't wait until Tuesday.

Here's what those six calls were worth. Using the per-call value range above ($350–1,200), a conservative blended figure of roughly £300 per won emergency job puts the weekend at about £1,800 in work — and that is an illustrative figure, not a measured one. The arithmetic: 6 emergency calls × ~£300 = ~£1,800. The point isn't the exact number. It's that every one of those six callers dialled the next firm on their list before you opened on Tuesday, and that work doesn't come back.

Why most of the money rings when you're closed

This isn't a staffing slip you can fix with one more person on the desk. It's structural. For service businesses, 25–40% of calls come after business hours — because pipes burst, boilers die, and ceilings leak on nights, weekends, and holidays, not at 11am on a Tuesday. Emergencies are, by definition, badly timed.

That flips the usual logic of a missed call. During business hours, a missed call is an annoyance. After hours, it's the whole game: the caller has a live emergency and a phone, and they will keep dialling until someone human-sounding picks up. Emergency calls outside business hours are revenue that goes to whoever answers first. If that isn't you, it's the competitor two listings down.

What does a missed call actually cost you?

Run your own version of the math — this is an illustrative worked example, not a quote.

Say you take 40 after-hours calls a month and currently answer none of them. Apply the documented per-call value range of $350–1,200 (industry estimates: Invoca 2024 pegs the home-services average at ~$1,200 per missed call; ServiceTitan data supports a $350+ floor for standard HVAC/plumbing jobs):

• At the low end: 40 × $350 = $14,000 in monthly opportunity walking out the door.
• At the high end: 40 × $1,200 = $48,000.

You won't win all 40 — some are tyre-kickers, some are wrong numbers. But you don't need all 40. Even recovering a quarter of the low-end figure is $3,500 a month in jobs you're currently handing to competitors for free. That single number is usually what makes the case for an after-hours answering service AI obvious — the leak is bigger than the fix.

(We don't take on projects under 500 calls/month, because below that volume the economics rarely clear. If your after-hours volume is well below that, a shared on-call human is probably the cheaper answer — and we'll tell you so.)

How an after-hours AI agent captures, qualifies, and dispatches

If you want the full mechanics of how a voice agent listens, understands intent, and acts, read the cornerstone explainer (in Russian). Here's the after-hours-specific version — the part that protects the £1,800 weekend.

1. It answers on the first ring, every time. No voicemail, no "leave a message and we'll call you back Tuesday." The caller with the flooded kitchen gets a live response at 2am.
2. It qualifies the job. What's the problem, is it an emergency or can it wait, what's the address, what's a callback number. The agent asks the questions your best dispatcher would ask.
3. It triages urgency. A burst pipe and a "can you quote me for a new boiler next week" are not the same call. The agent separates the genuine emergency from the routine enquiry.
4. It dispatches or schedules. A true emergency gets routed straight to your on-call technician — a phone handoff, an SMS, a Telegram card — with the job already qualified. A non-urgent enquiry gets captured as a clean lead for Tuesday morning, so nothing slips.

The technical bar for this is no longer the hard part. In 2026, response latency is sub-500ms (around 400ms) — the agent doesn't sound like it's thinking. That's table stakes now, not a selling point, which is exactly why the value lives in the qualify-and-route logic above, not in how the voice sounds.

The one question to settle before you start: can the AI reach your data?

Here is the lesson we paid for the hard way, and it has nothing to do with the voice.

We ran an after-hours voice operation for Pachamama Group in London — five restaurants, around 18,000 calls a month, effectively 24/7, on an ElevenLabs + n8n stack that pushed structured cards into Telegram. (That's a restaurant case, included here as proof that this scales to serious volume — not as a testimonial from a service business.) It worked. We wound it down anyway — not because the technology failed, but because OpenTable closed its API, and the agent could no longer reach the booking system it needed to do the job.

The takeaway for a service business: a voice agent is only as useful as its access to your dispatch or CRM system. Before we start, we run a data-access audit of your CRM and dispatch software. If the system you rely on is closed — no API, no integration path — we don't start. We'd rather tell you that up front than build something that can't actually book and route the jobs it catches.

Where the AI stops and a human takes over

An after-hours agent is not a replacement for your on-call technician, and we won't pretend otherwise. A true emergency still needs a human dispatched to the property — the AI doesn't fix the boiler. What it does is make sure the call is answered, the job is qualified, and the right human gets a clean, prioritised handoff instead of a voicemail discovered hours later. The agent catches and routes; people do the work. For the small share of calls that fall outside what it's been set up to handle, it escalates to a human rather than guessing.

One compliance line you can't skip

From 2 August 2026, the EU AI Act (Article 50) requires that callers be told they're speaking with an AI at first contact. We hard-code this as a deterministic opening line, every call, no exceptions — Article 50 transparency breaches carry fines up to €15M or 3% of turnover. We cover how disclosure is implemented in detail in our EU AI Act compliance guide (in Spanish). It's not a footnote; it's the first thing the agent says.

Setup, timeline, and what it costs

Setup starts from €3,000, with the managed service from €800/month plus per-minute usage. A typical build goes live in 14 days. For a service business losing four-figure weekends to a closed line, the question isn't really the setup fee — it's how many more Bank Holidays you want to hand to the firm two listings down.

See it answer your own line

The fastest way to judge an after-hours agent is to be the caller. Enter your company's URL and the AI agent will call you back in 3 minutes — you'll hear exactly how it answers, qualifies, and routes a job. Expect a call from your AI agent within 3 minutes.

Prefer to dial in yourself? Call the live demo line at +44 [TODO: data-point] and put it through its paces.

<!-- asset EN-3 · synthetic personas not used as testimonials · byline = Vladimir Nagin · keyword volumes pending Рита validation -->