Build an AI voice agent that qualifies leads by phone

An AI voice agent that qualifies leads can call each new lead within a minute of the form, ask your qualifying questions in a natural voice, and book the good ones on your calendar. ZeroCrew built this use case for founders who sell a service or a product with a sales call, and who can't drop everything each time a form comes in.

The problem with qualifying leads by phone is speed, not effort. ZeroCrew wrote this use case for the founder who collects leads from a website form, an ad, or a referral, then calls back hours later, after the lead has cooled or booked someone else. A voice agent that calls first, asks the same questions every time, and writes the answers into your CRM closes that gap.

ZeroCrew's version of this automation has four parts, a form, a voice agent, a calendar, and a CRM, joined by an automation tool. Below are the tools for each part, the prompt to paste, the setup steps, and the consent rules to follow before any agent dials a lead.

Why is qualifying leads by phone so hard by hand?

Qualifying leads by phone by hand is hard because leads arrive while you're working, and a slow reply loses them. Research by Oldroyd, McElheran, and Elkington, published in Harvard Business Review (2011) and archived by Brigham Young University, found that most companies don't respond to online leads nearly fast enough. ZeroCrew reads that finding as the case for a call that starts on its own.

Reaching a lead later is harder than it sounds. According to a Pew Research Center survey of 10,211 U.S. adults (July 2020), about eight in ten Americans say they generally don't answer cellphone calls from unknown numbers. ZeroCrew's takeaway for founders: call while the lead still expects your call, because a callback hours later looks like one more unknown number.

A qualifying call has more steps than people expect. ZeroCrew counts six in this use case: greet the lead, confirm they asked to be called, learn what they need, check budget and timing, book a meeting or say a kind no, and write it all down. Done by hand between client work, the notes are the step that slips, and the CRM ends up half empty.

How does an AI voice agent qualify leads?

An AI voice agent qualifies leads in five moves: trigger, call, qualify, book, and log. ZeroCrew wrote the chain below from the tools' own descriptions in its directory, opened in November 2026, and the whole chain runs without you opening an app:

1. A lead fills in your form and ticks a box agreeing to a call.
2. An automation sends the lead's name, number, and answers to the voice agent, which calls within a minute.
3. The agent confirms the lead asked for a call, then asks your qualifying questions: the need, the timeline, the budget, and who decides.
4. A lead who fits gets two open times from your calendar and books one. A lead who doesn't fit gets a kind answer and a useful next step.
5. The transcript, the answers, and a fit score land in your CRM, and you get a short summary.

An AI voice agent that qualifies leads can answer the phone too. ZeroCrew's AI voice agents page lists platforms such as Retell AI and Bland AI whose agents both answer and place calls, so a lead who rings your number gets the same questions as a lead who fills in the form.

Which tools do you need for an AI voice agent that qualifies leads?

An AI voice agent for lead qualification needs four kinds of tools: a form that captures consent, a voice agent that calls, a calendar that takes the booking, and a CRM that keeps the record. ZeroCrew ranks the alternatives for every step on its job pages, and the table below gives two options per step and when to pick which.

Step	Two options	Pick the first when	Pick the second when	Alternatives on ZeroCrew
Capture the lead	Tally or Typeform	You want a free form you write like a doc	You want one question at a time and booking through Calendly or Google Calendar	Feedback and surveys
Call and qualify, built your way	Retell AI or Vapi	You want a visual canvas and simulated test calls	A developer will configure it through the API	AI voice agents
Call and qualify, done with help	11x or Synthflow	You want a full AI sales team and will book a demo	You want no-code flows and a guided start	AI SDRs and AI receptionists
Book the meeting	Calendly or Cal.com	You want a free start and routing forms	You want open-source code you can self-host	Calendar
Keep the record	HubSpot or Attio	You want a free CRM with 2,000+ integrations	You want records enriched for you	CRM
Connect the steps	Zapier or Make	You want the widest app list	You want every step on a visual canvas	Automation

The voice agent is the step of lead qualification where founders choose most carefully. ZeroCrew's AI voice agents page ranks four platforms three ways: ElevenLabs is the most searched, solo founders say they use Vapi most, and Retell AI draws the fewest price complaints, as of November 2026. ZeroCrew's Vapi vs Retell AI comparison puts the two builders side by side.

Voice agent prices are per minute, so a short qualifying call costs little. According to each tool's own pricing page (as of November 2026), Retell AI charges $0.07 to $0.31 a minute with $10 in free credits, Bland AI's Start plan charges $0.14 a minute with no platform fee, and Vapi charges $0.05 a minute for hosting with $5 in free credits, plus transcription, model, and voice passed through at cost and phone costs set by your carrier. ZeroCrew links each one from its voice agents page.

A ready-made sales agent fits founders who want results before they want control. ZeroCrew's 11x page describes Julian, a voice agent that qualifies inbound leads and books meetings, with integrations for HubSpot, Salesforce, and Pipedrive. ZeroCrew's Synthflow page describes a visual Flow Designer and simulated test calls, and Synthflow starts with a demo, which suits a founder who wants help building without code.

The automation and CRM steps keep every call on file. ZeroCrew's Retell AI page lists built-in integrations with HubSpot, n8n, and Make, so a founder on those tools may need no extra connector. Everyone else can use Zapier, which connects more than 9,000 apps, or Make, or self-host n8n, all listed on ZeroCrew's Automation page.

You are the AI assistant for {registered business name}. You call new leads who asked for a call on our form, and you answer calls from leads who ring {business phone}. Your goal on every call is to find out whether the lead is a good fit and, if they are, book a meeting with {owner name}.

On an outbound call, start with: "Hi {lead name}, this is the AI assistant for {registered business name}. If you'd rather we didn't call, say stop at any time and I'll add you to our do-not-call list. This call is recorded and transcribed. You asked us to call you about {form topic}. Is now a good time for a few quick questions?" If it isn't, offer to call back at a time they choose, or text them {booking link}.

On an inbound call, start with: "Thanks for calling {registered business name}. I'm the AI assistant, and this call is recorded and transcribed. How can I help?"

1. Ask what they need help with and what made them reach out now.
2. Ask your qualifying questions one at a time, and listen before you ask the next:
   - What result do you want, and by when? A good fit needs it within {timeline}.
   - What budget have you set aside? A good fit has at least {minimum budget}.
   - Who else decides on this?
3. If the lead is a good fit, offer the next two open times from the calendar and book the one they pick. If you can't book on the call, text them {booking link}.
4. If the lead isn't a good fit, thank them, say kindly that {registered business name} isn't the right match today, and offer {helpful next step}.
5. Read back their name, number, email, and the time booked, and correct anything they change.
6. Answer questions only from this information: {services, prices, and FAQ}. For anything else, say {owner name} will answer it on the call.
7. If anyone asks not to be called again, say you'll remove their number, end the call politely, and mark them do-not-call.
8. Before you end every call, say they can reach {registered business name} at {business phone}.
9. Never pressure a lead, never promise a result, and never call anyone who didn't ask for a call.

After every call, send to {CRM} and {your email}: the lead's name and number, their answers to each question, a fit score from 1 to 5 with one sentence on why, the meeting booked or the next step, and any do-not-call request.

Paste the prompt into your voice agent's instructions, the box most platforms call the prompt, the agent's script, or its system message, and replace every placeholder with your own details. ZeroCrew wrote the prompt in plain sentences so it works on the voice agents in its directory, and you can add rules at the end as you learn from real calls.

How do you set up an AI voice agent to qualify leads?

Setting up an AI voice agent to qualify leads takes eight steps and about a day for a first version. ZeroCrew wrote each step from the tool descriptions in its directory, opened in November 2026; screens differ from tool to tool, so each tool's own help pages have the current version.

1. Write down what a good lead looks like: the need you solve, the budget you accept, the timeline you can meet, and who has to say yes.
2. Add a consent checkbox to your form in Tally or Typeform, with your company name and the phone number field, so each lead agrees in writing to a recorded call from an AI assistant.
3. Pick a voice agent such as Retell AI, Vapi, or Bland AI from ZeroCrew's AI voice agents page, start its free credits or free plan, and buy or connect a phone number.
4. Paste the prompt from this page and fill in every placeholder.
5. Connect your calendar, or give the agent your Calendly or Cal.com link to text to leads.
6. Connect the form to the agent with Zapier, Make, or n8n, so each new form entry starts an outbound call.
7. Send the transcript, answers, and fit score to HubSpot or Attio, and a summary to your email.
8. Fill in your own form five times and play five leads: a perfect fit, a lead with no budget, a lead in a hurry, a lead who asks if it's a robot, and a lead who asks not to be called.

What should you check before you trust an AI voice agent with leads?

Trust an AI voice agent with leads after a week of test calls and real calls you've read. ZeroCrew suggests reading every transcript in the first week, because a wrong answer about price, a booking at a bad time, or a missed opt-out shows up in the transcripts before it shows up in a complaint.

Consent comes first for any outbound call. The Federal Communications Commission ruled in February 2024 that AI-generated voices count as artificial under the Telephone Consumer Protection Act, and FCC rules at 47 CFR 64.1200 require prior express written consent before telemarketing calls with an artificial voice to a cellphone. ZeroCrew's use case calls only leads who ticked that box; ask a lawyer to review your form.

ZeroCrew's first-week checklist for an AI voice agent covers what most often goes wrong in a qualifying call, plus the rules in 47 CFR 64.1200 for every artificial-voice call:

• Every outbound call goes only to a lead who ticked the consent box, and only between 8 a.m. and 9 p.m. in the lead's time zone.
• The agent says it's an AI assistant for your company at the start of the call, says the call is recorded, and gives your callback number.
• On an outbound call, the agent offers a way to opt out right after naming your company.
• A lead who says "don't call me" is marked do-not-call in your CRM and never called again.
• Every booked meeting appears on your calendar with the lead's answers attached.
• The fit score in your CRM matches what you'd have decided after reading the transcript.

How does ZeroCrew help you choose an AI voice agent for leads?

ZeroCrew helps you choose an AI voice agent by ranking the options for every step, so your leads get called by an agent that fits your budget and your skills. ZeroCrew's AI voice agents page ranks the builders three ways, its AI SDRs page lists ready-made sales agents, and its CRM page ranks where the answers land.

Your own AI assistant can compare voice agents, forms, and CRMs with ZeroCrew's public MCP server at https://zerocrew.ai/api/mcp/public. The server has tools to search the directory, read a tool's page, list roles, and list the tools for a job, so Claude or ChatGPT can pull every voice agent, form, and CRM option before you pick. ZeroCrew's MCP servers list shows which tools in this chain have their own MCP server, and Retell AI, Vapi, Bland AI, and Synthflow all do.

For the inbound side of the same phone line, ZeroCrew's use case on answering every missed call with an AI receptionist shows how to book the job when a lead calls you. If you'd rather hand lead calls to someone, ZeroCrew's AI agent setup picks the agents, connects them to your tools, and keeps them running, starting at $1,000 a month.