# Inbound Call Automation: A Setup Guide

> Inbound call automation explained: how AI call handling answers incoming calls, after-hours setups, transfers, live data and what to measure after launch.

Source: https://business.mysmartassistant.ai/blog/inbound-call-automation · Published 2026-10-23 · Language: English · Publisher: MySA Business (https://business.mysmartassistant.ai)

Inbound call automation means letting software handle the calls customers make to your business, instead of a person picking up each one. With AI call handling, the caller talks normally and an AI voice answers questions from your approved information, books appointments, looks up live data, arranges callbacks and transfers the call to your team when it should. The aim is simple: no caller goes unanswered, and your staff handle only the calls that need them.

This guide is about planning inbound automation as a system: deciding which calls to automate, setting it up, covering after-hours, and measuring whether it works. For the front-desk view, see our guide to an [AI receptionist for business](/blog/ai-receptionist-for-business).

## Inbound call automation, then and now

Older ways to automate incoming calls were mostly about routing: an IVR ("press 1 for sales"), a call queue, a voicemail box. They decide who gets the call but do not answer the question. AI call handling goes further and resolves the call itself when the answer is in your information.

| Approach | What the caller experiences | What it resolves |
| --- | --- | --- |
| IVR menu | Key presses, then waiting | Routing only |
| Voicemail | A message, then waiting for a callback | Nothing until someone calls back |
| Human answering | A conversation, if someone is free | Everything, within working hours |
| AI call handling | A conversation in their language, any time | Routine questions, bookings, callbacks; transfers the rest |

## Step 1: Sort your incoming calls

Before switching anything on, take a week of calls and sort them into three groups:

- **Answerable from information.** Timings, prices, location, availability, documents needed.
- **Actions.** Book a slot, request a callback, check an order.
- **Needs a person.** Complaints, negotiations, unusual requests.

The first two groups are what you automate. The third is what call transfer is for. If most of your calls fall in the third group, automation will help less, and it is better to know that early.

## Step 2: Prepare the information the AI answers from

The AI answers only from the Business Details category you choose for the number. Write it as you would brief a new employee: prices with what they include, timings including holidays, what you do not offer, and the answers to the questions in your first group. Gaps here become "let me check" answers on the call, which is better than an invented answer, but still a missed chance.

For answers that change daily, connect your own API. With Custom APIs (enabled per workspace, set up by the Owner under Integrations → My APIs), you tick the APIs a number may use, and the AI can call them during the call when the customer mentions something, such as an order number or a repair ticket. An API failure never stops the call. See [Custom APIs](/docs/custom-apis).

## Step 3: Configure the number

In MySA Business, inbound is switched on per number you have bought in the app, so one number can be answered by the AI while another stays with your staff. On the number you set:

- **Business knowledge:** which Business Details category to use.
- **AI instructions:** what to do on these calls, up to 5,000 characters, or a saved purpose.
- **Language:** Hinglish by default; choose whether to stick to it or follow the caller.
- **Maximum call length:** caps what one call can cost.
- **Callbacks:** whether the AI may promise and place a call back when asked.

Inbound runs in your Live workspace and is charged from the same wallet as outbound calls; if the balance is empty, the call is not answered. The step-by-step is in the [inbound calls documentation](/docs/inbound-calls), and the overview is on [AI inbound call answering](/features/inbound-calls).

## Step 4: Decide your after-hours call answering

After-hours is where inbound automation often pays for itself, because nobody is there to answer otherwise. You have two choices:

1. **Answer around the clock.** Inbound is not blocked by default, so the AI answers night and day. Your team reviews summaries in the morning and rings back the warm leads.
2. **Set quiet hours.** If you prefer, [DND hours](/docs/dnd) can stop the AI answering during the window you set.

A common pattern is option 1 with appointments switched on: with [Appointments](/features/appointments) set up, a caller at 11 pm can book a slot for tomorrow and receive a WhatsApp confirmation, without anyone waking up.

## Step 5: Plan the handover to people

Automation that traps callers does more harm than a ringing phone. With [call transfer](/docs/call-transfer) ticked for inbound calls, a caller who asks for a person, or whose call matches a rule you wrote, is handed to your support number after the AI briefs your colleague. Transfer minutes are billed separately at the transfer rate shown in your workspace. Without transfer, the AI can still promise a callback if you allowed it.

### Example

> **Caller:** Mera AC service ka complaint pichhle hafte se pending hai, kisi se baat karwaiye.
> **AI:** Samajh gaya, main aapko abhi hamari service team se jod raha hoon. Ek second line par rahiye.
> *(The AI rings the support number, tells the colleague the caller's name and that a service complaint has been pending for a week, then connects the call.)*

## Step 6: Measure what changed

After the first two weeks, look at:

- How many calls were answered that previously went unanswered.
- How many ended in a booking, callback or Positive Lead.
- Which questions the AI could not answer; add them to your Business Details.
- How many calls were transferred, and why.

Every inbound call appears in call logs with a recording, an AI summary, sentiment and tags, with a direction filter to separate it from outbound. Turn-by-turn transcripts are not stored. Tags come from your own list, which makes [call tags](/features/call-tags) a useful way to count what callers ask about.

## Frequently asked questions

### What is inbound call automation?

It is handling incoming business calls with software instead of a person answering each one. With AI, the software holds the conversation, answers routine questions and passes the rest to your team.

### Which incoming calls should I automate first?

Calls answerable from written information, such as timings and prices, and simple actions like booking a slot or requesting a callback. Keep complaints and negotiations with your team through call transfer.

### Can I automate incoming calls only after office hours?

Inbound is switched on per number and answers whenever it is on. Many businesses use a separate number for the AI, or keep it answering around the clock and let staff ring back leads in the morning.

### Can the AI look up an order or booking during the call?

Yes, if your workspace has Custom APIs enabled and the API is attached to the number. The AI calls it when the customer mentions something it needs to look up.

### What happens if my wallet runs out?

Inbound calls are charged from your prepaid wallet. If the balance is empty, the call is not answered, so keep a buffer.

### Does AI call handling work in Hindi?

Yes. Calls default to Hinglish, and Hindi can be chosen as the call language for the number. You can let the AI follow the caller's language or stick to the one you chose.