Lyron
Communication · AI

AI Phone Assistant

An assistant that answers every call – at 9pm and when three lines ring at once. It books appointments, answers standard questions and hands over the moment it gets individual.

Context

The most expensive call is the one still ringing

In many businesses the phone is the one channel where being unreachable costs money instantly. Someone who cannot get through to a tradesperson, a practice or a law firm rarely calls twice – they dial the next number.

A classic answering machine does not fix this, it just moves the problem: the work still waits, only now as a callback list. An assistant picks up and closes the matter – or recognises that it cannot and transfers.

What we do not build: an assistant that negotiates prices, gives professional assessments or makes binding commitments. It answers, classifies, books what is bookable and hands over the rest with a complete note.

Use cases

Which calls are worth it

We start with the request type that occurs most often and has the clearest rules. Others follow once the first runs cleanly.

Biggest lever

Appointment booking

The use case with the clearest benefit: the assistant knows your free slots and books straight into the calendar – with confirmation by SMS or email.

New appointmentRescheduleCancelReminder

Recurring standard questions

Opening hours, directions, price ranges, required documents. Questions that come ten times a day and move nobody forward.

Opening hoursDirectionsDocumentsWho to ask

Calls outside business hours

Evenings, weekends, lunch break. Instead of a recorded message there is a captured case with a callback request and time window.

EveningWeekendPublic holidayLunch break

Absorbing call peaks

Monday morning, after a campaign, after an incident. The assistant answers in parallel instead of putting callers in a queue.

Parallel callsNo hold music

Qualify and transfer

The assistant establishes what it is about and connects to the right person – or creates the case for them.

Capture requestTransferCallback request

Order and status enquiries

Connected to your system it also answers questions about the specific case rather than only in general.

Order statusDelivery dateInvoice number
Example

One call, second by second

A conversation of 1:23 minutes: appointment booked, one question deliberately left open – and what the team receives from it.

Call log Duration 1:23 · Tue 29/07 · 20:14
Call answeredGreeting with company name and notice that this is a digital assistant.
Request identifiedAppointment wanted – heating system maintenance.
Availability checkedThree slots offered, travel time to the area taken into account.
Appointment bookedWednesday 6 August, 08:00 – confirmation sent by SMS.
Question outside scopeQuery about cost coverage – no binding answer given, callback promised.
Note and task createdCall note in the CRM, callback task with a deadline for inside sales.
Call note for the team
Caller
0231 · unknown number
Request
Heating maintenance appointment
Outcome
Booked Wed 06/08, 08:00
Channel
Confirmed by SMS
Open
Question about cost coverage
Next step
Callback by inside sales before 12:00

The amber line is the most important moment of the call – that is where a human takes over. We define exactly that boundary with you before the first call.

Illustrative transcript. The decisive moment is 1:02: instead of inventing an answer about cost coverage, the assistant promises a callback and creates a task with a deadline.

How it works

What the assistant does during a call

  • Answer and classify

    The assistant picks up after the first ring, states your company name and points out that it is a digital assistant. Then it establishes what the call is about.

  • Stay inside the scope

    For every request type it is defined in advance what it may do: book an appointment yes, quote a price only as a range, make commitments never. Anything outside that scope leads to a transfer.

  • Settle what can be settled

    Check the real calendar and book, answer a standard question, take a callback request with a time window. Not: say something so the call ends.

  • Hand over rather than guess

    On uncertainty, complaints or binding questions it transfers – or, outside business hours, promises a callback with a deadline.

  • Note it and pass it on

    Every call produces a structured note in the CRM: request, outcome, open points, next step. Including the calls that were transferred.

Impact

What changes about reachability

Today

  • Calls during an appointment go unanswered
  • The answering machine produces a callback list, not results
  • In the evening and at weekends nobody reaches anybody
  • During peaks several callers land in nothing at once
  • What was discussed sits on a note, or nowhere

With an assistant

  • Every call is answered, in parallel too
  • Appointments are booked instead of needing a callback
  • Outside hours a case with a deadline is created
  • Call peaks run through without a queue
  • Every conversation leaves a structured note in the CRM
Limits

Where a phone assistant has no business

On the phone, the damage from a wrong answer is greater than in any other channel. We fix these four boundaries before the first call:

  • No binding commitments. Prices only as a stored range, no discounts, no appointments outside the calendar, no professional assessments. Anyone asking is transferred or gets a callback – that is not a weakness, it is the point.
  • Emergencies go to a human immediately. For your industry we define keywords that transfer the call at once or state an emergency number. An assistant offering appointment slots during a burst pipe is a reputational risk.
  • Understanding has limits. Strong dialect, noisy surroundings, poor mobile reception – in those cases it transfers after two failed attempts rather than asking again. Nothing annoys callers more than a system asking them to repeat themselves three times.
  • It stays audibly an assistant. We do not build it to pass as human – partly because the EU AI Act requires transparency, partly because the disappointment when it is noticed outweighs anything gained by the pretence.
Systems

Runs with your telephony and your calendar

Microsoft 365Google CalendarCRM systemsSIP / phone systemSMS deliveryMicrosoft Teamsn8n
Scope and price

Scope and price

The entry price covers a production assistant with calendar integration. What moves the price, we say before the quote.

from €5,900 one-off
  • Set up a number or forward your existing one
  • Call script for up to three request types
  • Calendar integration with real availability checking
  • Escalation rules including emergency keywords
  • Call note handed over to your CRM or inbox
  • AI disclosure and EU AI Act documentation
  • Test operation with real calls before go-live

What increases the price

  • Several request types with their own scripts and follow-up questions
  • Connection to an industry or inventory system for status enquiries
  • Additional languages with their own quality review
  • Several locations with different hours and responsibilities
  • Connection to an existing phone system instead of a cloud number

Several request types with system integration and multiple locations typically land in the range of our Workflow Enterprise package from €5,900 upwards. We quote the binding fixed price after the test call.

All prices excl. VAT · telephony and model costs run through your own accounts

Included

What you get

  • Production phone assistant

    Reachable on your number, tested and signed off with real calls

  • Documented call scope

    What it answers, what it books, what it transfers – traceable and changeable

  • Note and handover flow

    Every call lands as a structured case where your team already works

  • Call analysis

    Which requests come how often, what gets transferred – the basis for the next stage

Questions & answers

Frequently asked questions about the AI phone assistant

Yes, and that is intended. The assistant says so in the greeting. Since 2 August 2026 the EU AI Act requires that people can tell when they are interacting with an AI system. Regardless of the obligation we consider it the better decision: someone who knows what they are talking to phrases things more clearly and is less annoyed by a transfer. Details in our guide to the EU AI Act labelling rules.
For your industry we store keywords that lift the call out of the normal flow immediately: either a direct transfer to an on-call number or a clear statement of which number to dial. That is the first thing we define with you – before any convenience feature.
Yes. The usual setup is forwarding: your number stays, and calls are passed to the assistant after a set number of rings or outside business hours. A human still answers when someone is there – the assistant only catches what would otherwise be lost.
Standard German with regional colouring works reliably; strong dialect and noisy surroundings remain difficult. That is why the abort is built in: after two failed attempts at understanding it transfers rather than asking again. During test operation we listen to real calls and tune it.
Three items: telephony minutes, speech and model costs, and optionally our support. At typical volumes the variable costs sit in the low double digits per month. We bill them transparently through your own accounts and do not mark them up.
No, and it should not be planned that way. It catches what is lost today: calls during appointments, in the evening, during peaks. If you have a staffed switchboard you gain relief – if you have none, you gain reachability. Both are different from replacement.

How many calls are you losing?

In the free intro call we work out which requests the assistant should take on and where the boundary runs. Then you hear a test call – before you decide.

Book a free intro call
Practical guide

Where AI phone assistant creates value in everyday work

A voice agent answers recurring calls, captures intent in a structured way, books suitable appointments and transfers sensitive conversations to your team.

Three concrete operating scenarios to compare with your own process.
01

Coordinate appointments

Availability is checked, customer details are captured and confirmations are sent immediately.

02

Pre-sort requests

Reason, urgency and callback details are handed to the right person in a structured format.

03

Extend availability

Standard questions can be handled reliably during peak demand or outside opening hours.

A strong fit when …

Recurring requests can be answered from approved knowledge while uncertain cases are handed to people with clear context.

  • You handle recurring calls using repeatable rules.
  • The intake, target system and accountable business role can be named clearly.
  • Exceptions are allowed to remain visible and move to people deliberately.
Transparent potential estimate

Estimate time savings with your own volume

The calculator uses 6 minutes today and 2 minutes after automation as fixed example assumptions. It does not replace process analysis.

Illustrative estimate based on the visible assumptions — not a guarantee.

43.3Hours per month
520Hours per year
Additional measures after launch Handling time Escalation rate Source coverage
Frequently asked questions

What decision-makers should know before starting

How does AI phone assistant work in practice?
A call arrives during, outside or in parallel with business hours. The workflow then validates the required data, runs approved steps and routes exceptions to the responsible person with context.
Which systems can be connected?
Typical integrations include Telefonanlage, Microsoft 365, Google Calendar, CRM, Teams. The decisive factors are a stable interface and clearly defined ownership of each data field, not a specific tool.
Which tasks deliberately stay with the team?
Emergencies, complaints, legally relevant statements and unclear requests are escalated to people immediately or after a defined clarification.
How is the automation introduced?
We start with a narrow knowledge and task scope, test real examples and expand only after documented approval. A tightly scoped first process typically takes 3–6 weeks; scope, interfaces and approvals determine the actual plan.
How can the benefit be measured?
Before implementation we record volume and current handling time. After launch we also compare Handling time, Escalation rate, Source coverage. The calculator on this page is a transparent estimate, not a promise.
Content reviewed on 26 July 2026 About Lyron AI