Most service businesses should start with one AI agent: the one that answers every call and message you currently miss. Once that’s working, add a follow-up agent and a CRM agent, in that order.
That order matters more than the tool you pick. Below are the seven agents we build for service businesses, what each one replaces, the one number that tells you it’s working, and how to choose your first.
Which AI agents do service businesses actually use?
Seven agents do almost all the useful work we see in service businesses. They fall into three groups, by where they sit in the business. For the full definition of an agent, read What is an AI agent?
For a longer list of real jobs, with examples from shops, installers and contractors, read AI agent examples. The seven below are the ones we start from.
- Voice agent (talks to customers): Answers every call when your team is busy or closed, answers routine questions, books the job during the call, and transfers emergencies to a person.
- Chat and DM agent (talks to customers): Replies to website chat, texts, Instagram and Facebook messages in seconds, answers from your approved information, and sends a booking link when the lead is ready.
- Speed-to-lead and follow-up agent (keeps the pipeline moving): Texts or calls a new lead within a minute, qualifies them, and keeps following up for days if they don’t book. Why speed matters is covered in Speed to lead.
- Qualification and CRM agent (keeps the pipeline moving): After every conversation, writes a two-line summary, tags the lead, fills fields like postal code and budget, and moves the deal to the right pipeline stage.
- Reactivation agent (keeps the pipeline moving): Starts personal, two-way text conversations with old leads and past customers and books the ones who reply. The full playbook is in Database reactivation.
- Review agent (runs the back office): Asks happy customers for a Google review after the job, replies to new reviews in your voice, and flags 3 stars or below to a manager. See How to get more Google reviews.
- Operations agent (runs the back office): Confirms and reschedules appointments, chases unpaid invoices, and answers staff questions from your SOPs.
What does each agent replace, and what should you measure?
Each agent replaces a specific manual job, and each has one number that tells you whether it’s paying off.
The time savings add up. In Thryv’s 2025 small business survey, 58% of owners already using AI said it saves them more than 20 hours a month, and 66% said it saves the business $500 to $2,000 a month.
| Agent | What it replaces | What to measure |
|---|---|---|
| Voice agent | Voicemail and “we’ll call you back” | Share of inbound calls answered (aim for 95%+) |
| Chat and DM agent | DMs answered the next morning, or never | Median first-response time (under 1 minute) |
| Speed-to-lead agent | Leads sitting in an inbox until someone has time | Lead-to-booked-appointment rate |
| CRM agent | Reps updating the CRM at 6 PM, or not at all | Share of leads with a complete record (100%) |
| Reactivation agent | A one-off email blast to an old list | Bookings per 100 contacts messaged |
| Review agent | Review requests nobody remembers to send | New Google reviews per month |
| Operations agent | An hour of reminder calls every morning | No-show rate and days-to-paid |
The first two agents matter most, because a missed call is usually a lost job. To put a dollar figure on that, read What is revenue leakage?
Which AI agent should your business start with?
Start with the agent that fixes your biggest leak. For most service businesses, that’s how they handle new leads, not what happens after the job.
Adoption is still uneven, which is the opportunity. In McKinsey’s 2026 State of AI survey, 40% of organizations with more than $1 billion in revenue said they were scaling AI agents. Among smaller organizations, the share stayed flat at 22%.
| Your business | Start with | Why |
|---|---|---|
| Owner-operated, under 20 leads a week (trades, solo clinics) | Voice agent + speed-to-lead agent | You’re losing leads while you’re on the job |
| A front desk or small sales team (dental, med spa, home services) | Chat and DM agent + CRM agent | Free the team from repeat messages and data entry |
| A big contact list (multi-location, high-ticket) | Reactivation agent + review agent | Revenue is already sitting in old leads and happy customers |
- Organizations over $1B revenue40%
- Smaller organizations22%
How do you launch your first agent in 30 days?
You can launch one agent in about a month if you keep the scope to one job.
- 1Week 1:Pick one job and one number. For example, “answer every after-hours call, measured by share of calls answered.
- 2Week 2Write the knowledge and the rules. Your 15 most common questions, the answers you want given, and exactly when the agent must hand off to a person.
- 3Week 3Connect the tools and test. Phone line, calendar, CRM. Run 20 or more test conversations, including the awkward ones.
- 4Week 4Go live, supervised. Read every transcript for two weeks, fix the gaps, then let it run.
PwC’s 2025 AI Agent Survey found that 42% of companies adopting AI agents were redesigning their processes around them. In a small business, that usually just means deciding who picks up a hot lead once the agent books it.
What stops AI agents from paying off?
Most failed agents fail for the same few reasons, and none of them is the AI model.
- One agent given five jobs at once, so it does none of them well.
- No guardrails, so it quotes prices or promises it shouldn’t.
- No clear handoff to a person, so frustrated customers have no way out.
- Not connected to the real calendar and CRM, so it “takes a message” instead of booking.
- Texts sent before A2P 10DLC registration in the US, or without consent under CASL in Canada. See A2P 10DLC explained.
- Measuring “conversations handled” instead of bookings and revenue.
Rule of thumb: if you can’t name the one number an agent is meant to move, it isn’t ready to launch.
Questions people ask
What are AI agents for business?
AI agents for business are AI systems given a goal and access to business tools such as the phone line, inbox, calendar and CRM, so they can take actions and not only answer questions. In a service business, they answer calls, follow up with leads, book jobs, and keep customer records up to date.
Which AI agent should a small service business start with?
A small service business should start with the AI agent that fixes its biggest revenue leak. For most owner-operated businesses, that means a voice agent that answers missed calls plus a speed-to-lead agent that replies to new form fills within a minute.
What is the difference between an AI agent and a chatbot?
A chatbot answers questions in a conversation. An AI agent has a goal and tools, so it also takes actions: booking appointments, updating CRM records, sending follow-ups, and routing urgent issues to a person.
How much does an AI agent cost for a small business?
The cost of an AI agent has three parts: the platform’s usage fees (per minute for voice, per message for text), phone and SMS charges, and the setup work. For voice agents specifically, Hexona’s guide to AI receptionist costs breaks down typical 2026 pricing.
How long does it take to launch an AI agent?
Launching one AI agent with one job takes about 30 days, including two weeks of supervised running. Most of that time goes into writing the knowledge base and the rules for when the agent hands off to a person.
Will AI agents replace staff in a small business?
In most small service businesses, AI agents don’t replace staff. They take over the repetitive first touch and admin work, such as answering routine calls and sending reminders, so the team spends its time on jobs and conversations that need a person.
Sources
Facts and prices checked on Nov 18, 2026. Prices change, so confirm on the source before you buy.
- McKinsey. 40% of organizations above $1B revenue scaling AI agents vs 22% of smaller organizations (1,719 respondents). Read Nov 18, 2026.
- Thryv. 58% of small businesses using AI save 20+ hours a month; 66% save $500 to $2,000 a month (540 decision-makers). Read Nov 18, 2026.
- PwC. 42% of companies adopting AI agents are redesigning processes around them (308 US executives). Read Nov 18, 2026.