An AI SDR, or AI sales development representative, is software that does the early stages of a sales development rep's job. It responds to new leads, asks qualification questions, handles common objections, books meetings on a rep's calendar and records everything in the CRM. Many AI SDRs work by email. An AI voice SDR does the same work on the phone, holding a live conversation with the prospect.
Voice suits the moment a lead is hottest, right after they fill in a form or ask for a demo, when a short call can confirm fit and lock in a meeting. This guide explains how an AI voice SDR works, where human SDRs still do better, how to design the qualification call, what goes wrong, the compliance rules that apply, and how to build one on ElevenAgents.
AI SDR vs Human SDR
An AI SDR handles repetitive, high-volume work well, while a human SDR is better at judgment and relationships. Most teams use both: the AI takes the first conversation, and a person takes over once there's a qualified opportunity.
| Task | AI SDR | Human SDR |
|---|---|---|
| Responding to a new lead within minutes, at any hour | Strong | Limited by working hours and workload |
| Asking a fixed set of qualification questions | Strong and consistent | Varies by rep and by day |
| Booking a meeting on a rep's calendar | Strong, if connected to calendars | Strong, but slower |
| Updating the CRM after every call | Automatic | Often delayed or incomplete |
| Handling common objections with approved answers | Good | Good |
| Handling unusual objections and pricing pushback | Weak | Strong |
| Reading company politics and deciding which accounts matter | Weak | Strong |
| Building a relationship with a key account | Weak | Strong |
Inbound vs Outbound AI SDRs
AI SDRs work in two directions, and each needs different preparation.
Inbound AI SDRs respond to people who have already shown interest, such as a demo request, a pricing form or an inbound call. The priority is speed: reaching the lead while they're still at their desk and still interested. Little list preparation is needed, because the prospect has already chosen to get in touch.
Outbound AI SDRs contact people who haven't engaged yet, working through a target list. The priority is reaching the right people at the right time, so most of the effort goes into building and cleaning the list, setting calling windows and retry rules, and confirming you have the consent to call.
Inbound is the easier and safer place to start. The prospect expects to hear from you, consent is simpler to establish, and the value of a fast response is well documented. In a study published in Harvard Business Review in 2011, firms that tried to contact a web lead within an hour were nearly seven times as likely to qualify it as firms that waited even an hour longer, and 23% of the companies audited never responded at all.
How an AI Voice SDR Qualifies a Lead and Books a Meeting
Here's how one inbound lead moves through an AI voice SDR.
The lead arrives. A prospect requests a demo at 7:40 p.m. Your form records their name, company, role and phone number, plus their consent to be contacted by phone.
The agent calls within minutes. It introduces itself, names your company, says it's an AI assistant, and mentions the demo request so the prospect knows why it's calling.
It checks what it already knows. Using a CRM lookup, it sees whether the company is an existing customer or an open opportunity, so it doesn't ask questions you already have answers to.
It qualifies. It asks a short set of questions about the problem, team size, current tools, timing and who else is involved, and adjusts based on the answers.
It handles objections. If the prospect says the timing is wrong or asks about price, it answers using approved information from the knowledge base, or offers to have a rep follow up.
It decides the next step. A qualified prospect gets a meeting booked on the right rep's calendar during the call, or a live transfer if they want to talk now. A prospect who isn't a fit gets a polite close and, where appropriate, a nurture email.
It writes everything to the CRM. The answers go into structured fields, with the transcript, a summary and the qualification decision, so the rep starts the meeting with full context.
A Qualification Call Template
The quality of an AI SDR depends on the conversation design more than on the model. This template works for most inbound B2B qualification calls. Adapt the questions and thresholds to your sales process.
1. Opening (under 20 seconds)
- Name the company and say the caller is an AI assistant
- Give the reason for the call, such as the demo request
- Ask permission to continue: "Is now a good time for two minutes?"
- If not, offer a callback time and end the call
2. Qualification questions (one at a time)
| Order | Question | What it tells you |
|---|---|---|
| 1 | "What made you look into this now?" | The problem and how urgent it is |
| 2 | "What are you using for this today?" | Current tools and switching context |
| 3 | "Roughly how many people would use it?" | Company and deal size |
| 4 | "When would you want something in place?" | Timeline |
| 5 | "Who else would be involved in the decision?" | Buying group and authority |
Keep it to five questions or fewer on the first call. A long interrogation over the phone loses people quickly.
3. Scoring
Give each answer a simple score and set a threshold in advance:
| Signal | Points |
|---|---|
| Clear problem that matches your product | 2 |
| Fits your target company size | 2 |
| Wants something within 3 months | 2 |
| Decision-maker or directly involved in the decision | 1 |
| Already using a competitor or a manual workaround | 1 |
For example, 6 or more points means book a meeting, 4 to 5 means offer a meeting and send information, and under 4 means close politely and add to nurture. The thresholds are yours to set, and you should adjust them after reviewing real calls.
4. Booking or transfer rules
- Book a meeting when the score meets the threshold and the prospect agrees
- Transfer live when the prospect asks to speak to someone now, or names a large deal or an urgent deadline
- Hand to a person for anything outside the agent's scope, such as custom pricing, legal questions or complaints
5. Fields written to the CRM
Problem, current tool, team size, timeline, decision role, score, qualification decision, meeting time or transfer outcome, call summary and transcript.
What AI SDRs Can Do Today: Examples From ElevenLabs Customers
ElevenLabs has published results from teams using ElevenAgents for sales development. These are ElevenLabs' reported figures, described in its guide to AI SDRs:
MyPlanAdvocate, an insurance brokerage, used AI voice agents to qualify Medicare-related inquiries. During its Annual Enrollment Period, the agent handled about 210,000 inbound calls a month, and calls passed to licensed representatives after AI pre-qualification converted at twice the historical rate.
ElevenLabs' own inbound SDR reviews contact form submissions, qualifies leads on the call, books meetings and records the decision in the CRM. ElevenLabs reports that 78% of its qualification decisions have needed no human intervention.
Employment Hero used an outbound voice agent to call existing customers about a new feature. ElevenLabs reports a 33% answer rate, with more than 3% of answered calls converting into activations.
Where AI SDRs Go Wrong
Thin qualification. An agent that books a meeting after one or two questions fills calendars with prospects who aren't ready. Reps then lose trust in every meeting the AI books.
Made-up answers. A language model can state something that sounds plausible and is wrong, such as a feature you don't offer or a price you don't charge. Limit the agent to approved knowledge and tell it what not to answer.
Poor data. Wrong phone numbers, outdated job titles and duplicate records lead to awkward calls and wasted minutes. Clean the list before any outbound campaign.
No clear handover. If the agent can't transfer smoothly, or the rep doesn't receive the call summary, the prospect has to repeat themselves and the AI's work is lost.
Calling without consent. In the US, calling a purchased list with an AI voice without the right consent breaks the TCPA. See the compliance section below.
Measuring the wrong thing. Counting meetings booked rewards the agent for booking weak meetings. Track meetings held and the pipeline they create.
Compliance for AI SDR Calls
This section is general information, not legal advice. Review your setup with your legal team before launch.
United States. Since a February 2024 FCC Declaratory Ruling, calls using AI-generated voices are treated as "artificial or prerecorded voice" calls under the Telephone Consumer Protection Act. They need the person's prior consent, must identify the business calling, and must offer an opt-out. Marketing calls carry stricter consent requirements than informational calls. For inbound AI SDRs, make sure your web form clearly states that people may be contacted by phone, including by automated or AI-generated calls, and keep a record of that consent.
European Union. Article 50 of the EU AI Act has applied since August 2, 2026. It requires that people be told when they're interacting with an AI system, so the agent should say it's an AI at the start of the call. GDPR applies to the call, the recording and the transcript.
India. Commercial calls fall under TRAI's Telecom Commercial Communications Customer Preference Regulations. Promotional calls must come from 140-series numbers and need valid consent, with Do Not Disturb preferences respected. The Digital Personal Data Protection Act, 2023 applies to the personal data you collect.
For more detail on outbound rules and spam prevention, see our guide to AI outbound calling.
Building an AI SDR on ElevenAgents
ElevenAgents provides the parts an AI voice SDR needs, according to ElevenLabs:
- Voices and models for natural, low-latency conversation, including Eleven v4 Turbo and Flash v2.5
- Workflows to keep the call in the right order: opening, qualification, scoring, then booking or transfer
- Knowledge base so the agent answers from approved product and pricing information
- Tools that call your systems during the conversation, such as a CRM lookup, a calendar availability check and a booking action
- Live transfer to a rep's phone number, with a call summary
- Voicemail detection for outbound campaigns
- Integrations with systems such as Salesforce, Zendesk and Twilio
- Choice of language model, so you can pick a model based on cost, speed or governance needs
- Testing and analytics to check calls before launch and review them afterward
A typical build follows these steps:
- Define the qualification rules. Agree on the questions, scoring, thresholds and handover rules with your sales team.
- Write the conversation. Turn the template into prompts and workflow stages, written for the ear: short sentences, one question at a time.
- Load approved knowledge. Add product, pricing and objection-handling content, and list topics the agent must not discuss.
- Connect tools. Set up the CRM lookup, calendar booking and transfer, starting with read-only lookups.
- Connect the trigger. For inbound, start a call when a form is submitted. For outbound, load a consented list with calling windows and retry limits.
- Test. Run test calls covering objections, interruptions, wrong numbers and people asking for a human.
- Launch small. Start with a share of leads, review every call in the first weeks, and adjust.
These are the planning estimates we use when we scope ElevenLabs projects:
| Scope | What's included | Typical timeline |
|---|---|---|
| Pilot | One inbound or outbound flow, a knowledge base and a CRM and calendar integration | 2 to 4 weeks |
| Production rollout | Inbound and outbound flows, live transfer, full CRM integration and a compliance review | 4 to 6 weeks |
| Multi-team or enterprise | Several agents, languages or regions, custom integrations and a security review | 8 to 12 weeks |
For the full process, see our guide to ElevenAgents implementation.
Metrics That Show Whether Your AI SDR Works
| Metric | What it tells you |
|---|---|
| Speed to lead | Time from form submission to first call |
| Contact rate | Share of leads the agent reaches |
| Qualification rate | Share of conversations that meet your threshold |
| Meetings held | Meetings that take place, not only those booked |
| Show rate | Meetings held divided by meetings booked |
| Rep acceptance | Share of AI-booked meetings that reps agree were qualified |
| Pipeline per meeting | Value of opportunities created from AI-booked meetings |
| Opt-out rate | Share of people asking not to be called again |
Rep acceptance and show rate tell you most about quality. If either falls, tighten the qualification questions or raise the threshold.
Frequently Asked Questions
What does an AI SDR do?
An AI SDR responds to new leads, asks qualification questions, handles common objections, books meetings on a rep's calendar and updates the CRM. An AI voice SDR does this through live phone conversations rather than email.
Can an AI SDR replace human SDRs?
It can take over repetitive work such as first responses, basic qualification, booking and CRM updates. Human SDRs remain better at complex objections, account strategy and building relationships, so most teams use both.
Is it legal for an AI SDR to call leads?
Yes, if you follow the rules where you call. In the US, AI voice calls need prior consent, caller identification and an opt-out under the TCPA. In the EU, the AI Act requires telling people they're speaking with an AI. In India, TRAI's rules on number series, consent and Do Not Disturb apply.
How long does it take to set up an AI SDR?
Our planning estimate for a pilot is 2 to 4 weeks, including qualification design, CRM and calendar integration, testing and a small launch.
Build Your AI SDR With Haass
Haass designs and deploys AI voice SDRs on ElevenLabs, from qualification design and CRM integration to compliance checks and rollout. If you're comparing partners, our guide to choosing an ElevenLabs implementation partner covers what to check.
See our ElevenLabs implementation services or book a call with our team.