AI Phone Calls That Work With Jobber: The Complete Guide to Inbound and Outbound Voice Agents

What is an AI voice agent for Jobber?

An AI voice agent is software that holds a natural spoken conversation on a real phone call, connected to your Jobber account so it can act, not just talk. On inbound calls it answers your number, greets the caller, looks them up in Jobber by their phone number, answers questions and books appointments into genuinely open slots. On outbound calls it dials customers for you: new lead callbacks, quote follow-ups, appointment confirmations and review requests after completed jobs.

The voice side is handled by a platform like Retell, which supplies the realistic voice, the listening and the conversation logic. The business side is handled by Jobber, which knows your customers, schedule, quotes and invoices. The craft is in connecting the two, and that connection is smaller than most people expect.

Why do phone calls decide who wins the job?

Because in home services, the phone is where urgency lives. A homeowner with a burst pipe or a dead furnace does not browse. They call, and they hire whoever picks up and can come soonest. Every call that hits voicemail is an invitation to call your competitor, and a large share of callers who reach voicemail never leave a message and never call back.

The math is brutal and simple. If an average job is worth several hundred to several thousand dollars, a handful of missed calls a week is a five-figure hole in the year. Hiring more office staff fixes it during office hours at office-staff prices. An AI voice agent fixes it at every hour for a fraction of that, and it never puts anyone on hold.

Texting, covered in part 1 of this series, catches the leads who type. Voice catches the ones who dial, and the ones who dial are usually the ones ready to buy now.

What can an AI voice agent do on inbound calls?

Answer instantly, know the caller, and finish the job the call was about. Here is what that looks like in practice.

Here is a condensed version of what a good inbound call sounds like:

What can an AI voice agent do on outbound calls?

Outbound is where voice AI quietly makes money, because every call type below is triggered by a Jobber event and runs without anyone dialing. One rule before the list: outbound automated calls need prior consent from the customer, and the section on etiquette below covers that.

Notice the pattern: the trigger is a Jobber event, exactly like the text-based scenarios in part 1. The only new ingredient is that the action is now "place a call" instead of "send a message."

How do you set up an AI voice agent in Retell?

Retell gives you the whole voice stack in one dashboard: the agent, the voice, the phone number, the transcripts. Here is the full journey in plain terms, no code required to get started.

1. Create your agent and give it a personality

   Sign up at Retell, create an agent and pick a voice from the library. Natural, warm and slightly slower tends to test best with older homeowners. Then write the agent's instructions, which are just plain English: who it is ("You are Riley, the scheduling assistant for Summit Plumbing"), what it must collect on every call (name, phone, address, what the job is, how urgent), how it should sound, and what it must never do (quote exact prices, promise arrival times, diagnose over the phone). Test it in the browser playground and keep refining until it sounds like your best office day.

2. Give the agent abilities

   Instructions make the agent talk. Abilities make it useful. In Retell these are functions the agent can use mid-conversation: check availability, book an appointment, transfer the call to a human, send a text, end the call politely. Each ability points at your bridge, which is the workflow that actually reads and writes Jobber. The agent decides when to use an ability the same way a human decides to check the calendar: when the conversation calls for it.

3. Attach a phone number for inbound

   Buy a number directly inside Retell, or connect the phone system you already have. Then assign your agent to answer that number. The smart rollout is conditional forwarding: your main line rings your office first, and only unanswered or after-hours calls forward to the agent's number. Your team notices nothing except that voicemail stopped filling up and the morning schedule started filling itself.

4. Trigger outbound calls from Jobber events

   Outbound calls in Retell are started by an instruction from outside: "call this number, as this agent, with this context." That instruction comes from your workflow tool. In Zapier, Make or n8n, you watch for the Jobber event (new request, quote quiet for three days, job completed) and fire the call with the customer's details filled in, so the agent opens with "Hi Sarah, you reached out about your gutters" instead of a cold hello. Always pass the context. A voice agent that knows why it is calling sounds brilliant, and one that does not sounds like spam.

5. Close the loop after every call

   When a call ends, Retell produces a transcript, a summary and the outcomes you asked it to track, such as "appointment booked" or "asked for a human callback." Send those to your bridge and write them into Jobber: a note on the client with the summary, the booking already on the schedule, or a task for your team when the agent promised a follow-up. This is the step that makes the whole system trustworthy, because anyone opening the client in Jobber sees exactly what was said and what happened.

How does the AI agent talk to Jobber?

Through a bridge, and the bridge is the same 3-step recipe from part 1 of this series. A Jobber app is created, you connect it to your Jobber account on a one-time approval screen, and the access and refresh tokens it receives let the bridge securely read and write your Jobber data. If you have not read that walkthrough, it is here: the complete Jobber automation guide.

With the connection in place, the bridge does four jobs for the voice agent:

• Recognize the caller. When a call comes in, look up the phone number in Jobber. If the client exists, the agent greets them by name and knows their property and history.
• Check real availability. When the caller wants a time, return only slots that are genuinely open on your Jobber schedule.
• Create what was agreed. New client, request, booking or reschedule, created in Jobber during the call, with duplicate prevention by phone and email.
• Record the conversation's outcome. After the call, attach the summary and result to the client in Jobber so the paper trail is complete.

The bridge itself can be a workflow in Zapier, Make or n8n for straightforward setups, or a small custom app when you want tighter logic. Ready-made AI platforms for home services ship with this bridge already built.

What are the rules for AI calls people actually like?

Voice is more intimate than text, so the etiquette bar is higher. These rules are the difference between "wow" and "spam."

• Say it is an AI, first sentence, every call. Disclosure is legally required for automated and recorded calls in a growing number of places, and hiding it destroys trust the moment it is discovered. Callers forgive an honest robot. They do not forgive a fake human.
• Keep a human one sentence away. "Let me get someone from the team" should work at any moment, instantly, on every call.
• Get consent before outbound. Automated outbound calls require the customer's prior consent in many jurisdictions, including under US telemarketing rules. Collect consent on your forms and in Jobber, honor do-not-call requests permanently, and keep calls to reasonable local hours.
• Know when not to call. An invoice three days overdue deserves a text, not a phone call. Save voice for moments where a conversation adds value: urgency, scheduling, quotes, thank-yous. The text side has its own playbook in part 3 of this series.
• Keep it short. The best outbound AI calls are under two minutes. State why you are calling in the first breath, do the one thing, get off the phone.
• Log everything in Jobber. Every call, transcript and outcome on the client record. When a customer says "your assistant told me Thursday," anyone on your team can see exactly what was said.

Should you build AI calling yourself or use a platform?

The same three honest paths as part 1, with one difference: voice has less room for "good enough." A clunky text message is ignored. A clunky phone call is hung up on and remembered.

1. Retell plus a workflow tool

Retell's dashboard plus Zapier, Make or n8n gets a basic inbound agent live without code, and it is the best way to learn what voice AI can do for you. The assembly work is real: prompts to refine, abilities to wire, transcripts to review, edge cases to handle.

2. Retell plus a developer

A custom bridge gives you live availability checks, watertight duplicate prevention and exactly your business logic. Right choice if you have development capacity and unusual workflows.

3. A ready-made AI platform for home services

Platforms built for this industry ship with the voice agent, the Jobber connection, the booking logic and the compliance guardrails already assembled and tested across thousands of calls. You configure your services, hours and voice, and connect Jobber in minutes. The fastest path to the outcome, and the only one where transcripts, texting and calling live in one place.

Frequently asked questions about AI phone calls and Jobber