Can an AI agent make phone calls?
Key Facts
- 85% of people whose calls go unanswered never call back, according to Retell AI.
- Businesses collectively lose over £30 billion every year to missed calls, Retell AI research shows.
- The global AI voice agents market will grow from $2.5 billion in 2025 to $35.2 billion by 2033, per Grand View Research.
- Inbound AI voice agents held a 52.1% revenue share in 2025, while outbound calling is the fastest-growing segment, Grand View Research reports.
- The FCC ruled in February 2024 that AI-generated voices fall under TCPA robocall rules, with penalties of $500 to $1,500 per violation, per Decagon.
- NiCE's AI agents resolve over one billion customer requests yearly across 100+ languages with roughly one-second response times, NiCE reports.
- Zoom warns the human handoff is the weakest point of most AI voice agent deployments, according to Zoom's analysis.
The Cost of the Call You Missed
It's 7:40 p.m. on a Tuesday. A homeowner with a burst pipe calls the first plumber she finds. Nobody answers. She dials the next listing, and that's the moment your business quietly lost a job you'll never know existed.
This is the daily reality for local service businesses. Calls arrive at dinner time, on weekends, while you're under a sink or in a courtroom. According to Retell AI, 85% of people whose calls go unanswered will not call back. They don't leave a voicemail, and they don't try you again tomorrow. They call your competitor.
The scale of the problem is bigger than any single missed job. The same research puts collective business losses from missed calls at over £30 billion every year. For an individual firm, the damage is invisible by design: the phone simply didn't ring on your end, so nothing shows up in your reports as a loss. You only count the jobs you won, never the ones that evaporated in the first four rings.
The math gets worse when you consider what those missed calls actually were. A missed call to a law firm is often someone in genuine distress, ready to hire. A missed call to an HVAC company during a heat wave is an emergency booking with full price acceptance. These are the highest-intent, highest-value leads a service business gets, and they're disproportionately likely to arrive outside office hours, when nobody's at the desk.
Consider what's really at stake when a call goes to voicemail:
- The caller rarely leaves a message, so you get no name, number, or callback opportunity.
- Most callers move straight to the next business on their list within minutes.
- Urgent jobs, the ones with the least price sensitivity, are the most likely to be lost.
- Repeat and referral business can vanish too, since a first impression of "nobody answers" sticks.
Some owners try to patch the gap with an answering service or a second receptionist shift, which helps but adds real payroll cost to hours when call volume is unpredictable. Others route after-hours calls to voicemail and accept the losses as the cost of doing business. Neither option changes the underlying problem: your availability and your customers' availability don't match, and the mismatch is expensive.
Here's the irony. As NiCE points out, the phone remains the channel customers choose for urgent, complex, and emotional issues, yet it has historically been the most expensive and hardest channel to automate well. The one channel your customers trust most when something breaks at 9 p.m. is the one your business can least afford to staff around the clock. That contradiction is exactly what modern tools like an AI receptionist set out to resolve, and understanding how Answrr works starts with asking why the phone was ever so hard to cover in the first place.
Yes: AI Agents Make Phone Calls Both Directions
Yes, an AI agent can make phone calls, and it's already happening in both directions at scale. AI phone agents are no longer experimental: they answer support lines, run outbound sales campaigns, handle renewals and collections, and often do it in multiple languages, around the clock. Teams now use them to answer inbound calls, qualify leads, book appointments, follow up with prospects, and hand conversations off to a human when needed, according to a Plivo analysis of AI voice agents for call centers.
The market data backs this up. The global AI voice agents market was worth USD 2.5 billion in 2025 and is projected to reach USD 35.2 billion by 2033, growing at a 39.0% CAGR, according to Grand View Research. The market is segmented by call direction, and both sides are thriving:
- Inbound voice agents held a 52.1% revenue share in 2025, handling support requests, inquiries, and service bookings.
- Outbound voice agents are the fastest-growing segment from 2026 to 2033, powering appointment reminders, payment follow-ups, and lead qualification.
- Modern agents take action mid-call: checking availability, booking appointments, and updating CRM systems in real time.
In practice, this means the same AI that answers your phone can also place calls on your behalf. Decagon notes that outbound AI handles proactive customer communication at scale, not just inbound answering. For a small business, that combination covers the two jobs a receptionist does: picking up every incoming call and making follow-up calls so leads and appointments don't slip through the cracks.
One caveat worth knowing: outbound AI calling is regulated. The FCC's February 2024 ruling confirmed that AI-generated voices fall under the TCPA, so AI-initiated calls require the same prior express consent as robocalls, with penalties of $500 to $1,500 per violation. Inbound answering carries none of that friction, which is why it remains the larger segment today.
For local service businesses, the practical version of this capability is an AI receptionist that answers every call 24/7, books appointments, and transfers urgent calls to you with full context. That's what we built Answrr to do, with white-glove setup so you don't have to configure anything yourself. If you want to see the mechanics, how Answrr works walks through the whole process. The AI phone system that sounds human and books appointments 24/7 is no longer a demo. It's answering real phones right now.
What the AI Actually Does on the Line
Can an AI agent make phone calls? Yes, and the more useful question is what happens after the line connects. A voice agent that only answers questions is a demo; one that books the appointment is a deployment.
That difference matters to businesses losing leads to voicemail. According to Retell AI's review of answering services, 85% of people whose calls go unanswered will not call back, and businesses collectively lose over £30 billion every year to missed calls.
Mid-call, a modern agent takes action rather than collecting a message. As Retell AI describes it, the agent checks availability, books appointments, updates a CRM or scheduling system, and captures caller information in real time.
- Check the calendar and offer genuine openings
- Book the appointment and confirm the details
- Qualify the caller against your criteria
- Write the caller details back to the CRM or scheduling system
Three components do the work, and NiCE describes the pipeline as automatic speech recognition, LLM reasoning, and neural text to speech. Recognition is the ear, the language model is the brain, and neural synthesis is the voice. NiCE reports its agents resolving more than a billion customer requests a year across over 100 languages, with responses landing in roughly one second.
The same pipeline places outbound calls for appointment reminders, payment follow-up, and lead qualification, which is why the market now divides into inbound and outbound voice agents, with outbound the fastest-growing of the two.
That is what separates an agent from a rigid menu IVR. A menu walks a fixed tree of options and dead-ends anything unusual, while an agent reasons about the goal in front of it, so "can I move my appointment" and "is anyone free before Friday" can both end in an action.
The hard part is the moment a call leaves the agent. Retell AI describes a warm transfer with full conversation context</strong
Outbound Calling Rules You Can't Ignore
Before your AI agent dials a single number, know this: the FCC's February 2024 Declaratory Ruling (FCC 24-17) treats AI-generated voices as "artificial or prerecorded" under the TCPA, even when the call is fully interactive. That means neural text-to-speech and voice cloning fall under the same rules as robocalls, according to legal analysis of the ruling. The law doesn't care whether a human or a script is talking. It cares about consent, disclosure, and behavior.
Consent is the first gate. AI-initiated calls require the same prior express consent as robocalls, and telemarketing calls using an artificial voice generally require prior express written consent. The price of skipping that step is steep: TCPA penalties run $500 to $1,500 per violation, with no cap on total damages.
A few operational rules apply to every outbound campaign:
- Honor opt-out requests within 10 business days.
- Keep calls inside the 8 a.m. to 9 p.m. local window for the person you're calling.
- Confirm you hold the right type of consent before any telemarketing dial.
- Review your disclosure language so callers know they're speaking with an AI agent.
Some platforms go further and require KYC verification before enabling outbound calling at all, which tells you how seriously the industry reads these obligations. Automation doesn't shrink your legal exposure. If anything, it raises the stakes because mistakes scale with every dial.
This is why consent records and disclosure scripts deserve a check before your first outbound campaign, not after. The same discipline applies whether you run an AI receptionist on inbound calls or add outbound follow-ups later. If you want to see how the setup handles both directions, how Answrr works walks through the configuration, and our team builds it with you rather than handing you a login and wishing you luck. The AI phone system that sounds human and books appointments 24/7 is worth having. Just make sure the calls it places are ones you're allowed to make.
Putting an AI Phone System to Work in Your Business
A plan beats a demo every time. The businesses that get real value from an AI phone system map their call flows before the first call goes live, then decide where automation stops and a person takes over.
Start with your top call types. List the reasons people actually phone you (booking, rescheduling, pricing questions, urgent issues) and rank them by volume. From there, mark which calls the AI handles start to finish and which escalate. That split matters: Zoom's analysis of voice agent deployments notes that the handoff to a human is the weakest point of most rollouts, and a poor escalation can undo the goodwill of a smooth automated conversation.
Next, connect your tools. An agent that can check availability, book the appointment, and update your CRM mid-call delivers far more than one that only takes a message. Then write your escalation rules: who gets an urgent call, how fast, and what context travels with it. Warm transfers that pass the conversation history forward are standard among leading services, so your team never asks a caller to repeat themselves.
Before any outbound campaign runs, verify consent. The FCC's February 2024 ruling confirmed that AI-generated voices fall under the TCPA's "artificial or prerecorded voice" rules, meaning AI-initiated calls need the same prior express consent as robocalls, with penalties of $500 to $1,500 per violation and no cap on total damages.
A white-glove setup handles the wiring for you. Answrr's team maps your flows, connects scheduling and CRM, configures escalation, and verifies outbound consent before launch. You can read how Answrr works for the setup details.
The payoff is an AI receptionist that sounds human and books appointments 24/7, with urgent calls transferred to your team and transcripts and summaries logged afterward.
Frequently Asked Questions
Can an AI agent actually make outbound phone calls, or does it only answer incoming ones?
What does an AI phone agent actually do during a call besides talking?
Is it legal for an AI to place outbound calls on my behalf?
What happens when the AI can't handle a caller's problem?
How many customers actually call back after reaching voicemail?
Do I need to disclose that callers are speaking with an AI?
Your Phone Can Now Work the Hours You Can't
So, can an AI agent make phone calls? The answer is a clear yes, in both directions. AI voice agents answer inbound calls, book appointments, qualify leads, and place outbound follow-ups, all in production today, backed by a market projected to grow from USD 2.5 billion in 2025 to USD 35.2 billion by 2033, according to Grand View Research. The stakes for local service businesses are just as clear: 85% of callers who reach voicemail never call back, and those missed calls are disproportionately your highest-intent, highest-value leads. The path forward is practical, not futuristic. Map your call types, connect your scheduling and CRM tools, define escalation rules with warm transfers, and verify consent before any outbound campaign. If you're ready to see how that looks in practice, how Answrr works walks through the full setup, and our team builds it with you, no DIY configuration required. Because the question was never whether an AI receptionist can pick up the phone. It's whether you can afford to keep letting it ring out.