AI Phone Agents for Dealerships: The 2026 Buyer's Guide

The average dealership misses roughly a third of its incoming calls. STELLA Automotive AI puts it at 33%. Car Wars’ 2025 data across thousands of stores shows a 63.5% connect rate. On the sales side, every missed call from a customer asking about a vehicle is a lead that went to the dealer who picked up. On 150 internet leads per month at a 12% close rate, even a modest improvement from answering more calls adds up fast. At $3,200 front gross, two extra deals a month is $76,800 a year. On the service side, Flai’s analysis of 600+ franchise stores found the average dealer misses 158 appointment-related calls per month. At a $450 average repair order, even if half call back or book online, that’s over $400,000 in annual service revenue at risk.

AI phone agents have tried to solve the missed-call problem. But 64% of customers prefer companies didn’t use AI for customer service (Gartner), and the ones who get handed off to a human who fumbles the conversation are twice as likely to walk (Pied Piper 2026). Getting the phone answered is step one. What happens on the call is where the deal is made or lost.

Most stores have tried hiring more people for the phones, but BDC turnover is notoriously high and every new hire takes 60-90 days to get competent. Overflow routing helps until the calls cluster at the worst possible times: Monday morning, Saturday afternoon, 10 AM to noon when the service drive is slammed. The staffing math doesn’t work. It never has.

That’s why AI phone agents exist. And that’s why eight companies have raised over $580 million in disclosed funding to build them. The question isn’t whether this technology works. It’s which provider fits your store, what it actually costs, and whether the humans on the other end of the handoff are any good.

What AI Phone Agents Actually Do (And Don’t Do)

An AI phone agent answers your dealership’s phone, handles basic inquiries, and routes complex calls to humans. In 2026, the best ones can:

• Answer every call instantly, 24/7, including after hours when 40% of customer inquiries arrive (Podium/OpenAI case study)
• Schedule service appointments directly in your DMS scheduler, understanding 270+ service opcodes (STELLA)
• Handle multiple calls simultaneously without hold times. Car Wars data shows the average hold time is 3 minutes 5 seconds. AI eliminates that.
• Get sales leads on the phone by asking about vehicle interest, trade-in details, and financing needs
• Run outbound recall campaigns and re-engagement calls (Numa, Flai)
• Transfer to a human with context when the conversation exceeds the AI’s capability

What they can’t do:

• Negotiate. Out-the-door pricing, negative equity conversations, payment structuring. These require a human who can read the room and work the desk.
• Handle emotional situations. The customer whose car was totaled. The service customer whose repair wasn’t done right. AI saying “I understand your frustration” makes it worse.
• Close deals. AI gets the phone answered. It doesn’t get the customer to say yes. That takes a salesperson who asks the right questions, handles the objection, and lands the payment.
• Coach. AI doesn’t hear the salesperson miss a buying signal, dodge the negative equity question, or talk for 9 minutes without asking for the appointment.

But the alternative to AI isn’t a perfect human interaction. The alternative is a phone that rings six times and goes to voicemail. The answer isn’t AI OR people. It’s a platform that gets the lead to a person fast and then makes sure that conversation counts.

The 8 Providers: Who Does What

Eight companies are competing for AI phone agent dollars in automotive. What each actually does, who backs them, and where they’re strongest.

Tier 1: Established Players

Numa raised $55M from Google, Threshold Ventures (early Tesla investor), and Touring Capital. They serve 1,200+ dealerships and have OEM approval from GM (iMR co-op reimbursement) and Stellantis (MarketCenter). In January 2026, they launched four products: Operator (AI receptionist), Opportunities (outbound AI), LiveCSI (real-time sentiment monitoring), and a unified communications platform. Numa is the only provider attempting real-time customer sentiment scoring. They’re an Inc. 5000 company (#168 fastest-growing).

STELLA Automotive AI raised $34.7M, with Reynolds and Reynolds as a strategic investor. Their AI understands 270+ service opcodes, handles unlimited concurrent calls, and publishes a NADA evaluation guide for dealers comparing AI providers. The Reynolds connection implies deep DMS integration for stores already on that system.

Podium (Jerry 2.0) is the largest player by funding at $440M total and a $3B valuation. Jerry 2.0 launched January 2026, powered by OpenAI’s GPT models. It serves 10,000+ businesses (not all automotive). The OpenAI case study reports ~30% revenue lift, 45% better lead close rates, and ~2-minute response times versus 2+ hours for humans. Podium hit $100M ARR in 21 months with Jerry. Their “thought process” transparency feature lets managers and customers see which policies and data the AI checked before answering.

Tier 2: Fast-Growing Challengers

Mia Labs raised $29M (Series A: January 2026). 350+ franchise dealerships. 1 million+ customer conversations, 130,000+ appointments booked, $45M+ in revenue from AI-booked appointments. Car Dealership Guy (Yossi Levi) is a strategic investor. The team has 100+ years combined automotive experience including dealership ops and OEM roles.

Toma raised $17M led by a16z. Y Combinator graduate. 100+ dealerships. Their differentiator is RLHF learning (“Toma IQ”) where handoff calls get analyzed to improve the AI specifically for each dealership’s patterns. SOC 2 Type II certified with zero exceptions. When Toma called every dealership in the country as part of their research, calls were only picked up 45% of the time.

Pam AI raised an undisclosed amount from Autotech Ventures, with backing from HF0 (backers of Ramp) and First Harmonic. They serve hundreds of dealerships across 33 states and 3 Canadian provinces. The broadest DMS integration list in the space: CDK, Xtime, WiAdvisor, DealerFX, MyKaarma, TCC, AutoLoop, RedCap, Auto.live, and Tekion. They report $50,000/month in service revenue per dealership on average and 95% retention. Named after Pam Beesly from The Office.

Tier 3: Specialized / Newer

Flai raised $6M from First Round Capital and Toyota Ventures. Y Combinator graduate. Built their own voice infrastructure from scratch rather than stitching together off-the-shelf components. Claims $80,000-$100,000+ monthly profit impact per store based on a Lexus Bay Area case study (premium brand, high-volume market, not typical). SOC 2 certified.

Brooke.ai (now part of Better Car People) serves 400+ stores and made a deliberate choice: service departments only. No sales. They integrate with Xtime and other schedulers. If your problem is specifically service appointment calls, Brooke’s focused approach means fewer distractions in the product.

Feature Comparison Matrix

“Unknown” means the provider has not publicly confirmed or denied this capability. It does not mean “no.” Feature info is based on public marketing materials and press releases as of April 2026. Capabilities change fast. Verify during your demo. Ringlead is not an AI phone agent. It’s the layer that makes the human side of the handoff work.

Want AI that does something useful for managers? Try the live demo and see how Ringlead connects leads, scores calls, and flags deals that need attention.

The gap shows up in the first four rows. Every provider answers the phone. None of them make the salesperson on the other end any better. More on that below.

What Dealers Are Actually Saying

On DealerRefresh, one dealer posted about building an AI agent to answer phones: “Yes, it answers basic questions, schedules appointments, or even qualifies leads. It’ll even answer complex questions, really handy for service.” The thread shows operators open to experimenting but cautious about customer perception.

On Reddit, a customer who called a dealership with an AI receptionist said: “Worked pretty good honestly. Not as good as a real person.” Mixed reviews from dealers: “Some dealerships say it’s helping them respond to leads faster and handle calls. Others say it’s just another layer.”

One UK automotive deployment on Reddit revealed the real bottleneck: “The DMS integration was 80% of the work. The voice AI part was straightforward.”

Named dealer results from provider press:

• Peter Bartlett, CTO, Hennessy Auto Group (Pam): “We’ve eliminated the need for a full-time receptionist, or more accurately, three of them.”
• Andrew Reagan, Southtown Auto (Podium): Had 4,000-4,500 monthly leads across 2 rooftops with 6 people. After Jerry, time-to-lead cut in half.
• Tina Tasche, Van Horn Auto Group (Podium): “The sweet spot is just having AI be the leading point, and then humans take over when needed.”

The Top 4 Objections From Your Team

1. “Customers will hate talking to a robot.” They’re not wrong. 64% of customers prefer no AI (Gartner). But 33% of calls go unanswered today, and silence is worse. The real play is AI catching the call and getting it to a human fast. The customer ends up talking to a person. That’s where the satisfaction comes from.
2. “What if it says something wrong about our inventory?” A real risk. Podium’s “thought process” feature shows what the AI checked. Others use guardrails and escalation rules. Ask every provider how they handle wrong-answer scenarios.
3. “My salespeople will resist it.” From DealerRefresh: sales teams see AI as a threat to their commissions on leads they would’ve handled. Frame it as catching missed calls, not replacing salespeople.
4. “DMS integration is the real problem.” The Reddit crowd confirms it. DMS integration should be the #1 evaluation criterion, not voice quality.

Pricing: What You’ll Actually Pay

Most providers don’t publish pricing. All require a demo and quote. Based on available data:

For comparison: the marginal cost of an AI voice call runs $0.30-$0.50 versus $12 for a human-handled call (Master of Code), though that doesn’t include the platform subscription. A 4-person BDC runs $150,000-$200,000 per year fully loaded with salary, benefits, management overhead, and chronic turnover.

6 Hidden Costs to Watch

1. DMS/CRM integration may require a one-time setup fee or ongoing connector cost.
2. Training AI on your specific inventory, promotions, and policies takes 1-4 weeks. That’s time your team is managing the transition.
3. Multi-year contracts with auto-renewal are common. Ask about early termination before you sign.
4. Multi-rooftop groups multiply per-location costs fast. Ask about group discounts.
5. Usage-based pricing models can spike during high-volume months.
6. If the provider offers live agent fallback when AI can’t handle a call, that may cost extra.

ROI Claims: Take With Salt

The Scoring Gap: Who’s Watching the Humans?

Here’s the scenario playing out at thousands of stores right now.

AI answers the phone at 7:48 PM. Qualifies the lead. Customer has a 2020 Civic trade, interested in a Tucson she saw online, pre-approved through her credit union. Books a test drive for 11 AM Tuesday. Good lead. AI did its job.

Monday afternoon, the salesperson calls to confirm. Really they’re qualifying, tightening up the appointment, getting trade details the AI might have missed. Customer picks up. She’s interested but tough. “I’m still comparing a couple options.” “What’s my trade actually worth?” “The payment on your website doesn’t include my negative equity.” The salesperson talks for four minutes. Doesn’t address the negative equity. Doesn’t offer to get a real appraisal on the trade before she comes in. Doesn’t lock down a specific vehicle or create any urgency to keep Tuesday. Says “sounds good, we’ll see you tomorrow” and hangs up.

Customer no-shows Tuesday. Salesperson doesn’t log the call. CRM shows “no show.” Manager pulls the report, sees another AI-booked appointment that didn’t convert, and starts questioning whether the AI is booking real customers or tire-kickers.

Nobody heard the four-minute call. No recording. It was from a personal cell. No score. No notes. The only thing in the CRM is “no show.” The AI booked a qualified buyer. The salesperson lost her on the phone and nobody will ever know.

Pied Piper’s 2026 ILE study backs this up: when AI encounters a situation requiring human help, satisfaction scores drop 9 points on average. Customers in those moments were twice as likely to receive no meaningful personal response at all. The AI set an expectation of speed and competence. The human interaction didn’t match.

Every provider on this list answers the phone. None of them make the salesperson on the other end any better. The hybrid AI+human model outperforms pure AI by 15-25% (Calldrip). Nobody’s listening to those conversations. That’s the problem.

None of the eight providers in this guide score call quality. Numa’s LiveCSI monitors customer sentiment but doesn’t coach calls. Toma’s IQ learns from handoff failures but doesn’t catch the salesperson who fumbled the negative equity question. The scoring gap is the biggest blind spot in the market.

That’s the gap we built Ringlead to fill. Speed-to-lead connects incoming leads to a live salesperson by calling every cell phone on the team simultaneously. During business hours, the customer is talking to a real person in under 60 seconds. After hours, they get instant SMS and email. Call recording captures every outbound call, including personal cell phones. The four-minute call that lost Tuesday’s appointment gets heard. AI call scoring grades every conversation A-F and surfaces coaching moments before the morning meeting.

If you’re also evaluating AI tools for the follow-up after the first call, our AI follow-up breakdown covers 10 providers separately.

20 Questions to Ask Before You Buy

Integration and Operations

1. Which DMS systems do you integrate with? (CDK, Reynolds, Tekion, DealerSocket, specifically)
2. Which schedulers do you connect to? (Xtime, MyKaarma, DealerFX)
3. Does the AI log calls and appointments directly into my CRM?
4. What’s the setup timeline? Days? Weeks?
5. Can I customize responses for my specific inventory, promotions, and policies?

Call Handling and Quality

6. What happens when the AI can’t handle a call? Warm transfer? Cold transfer? Voicemail?
7. What’s the average transfer time to a human when the AI escalates?
8. Can I listen to every AI-handled call?
9. Can I score AI-handled calls on the same rubric as human calls?
10. How does the AI handle angry or upset customers?
11. What’s the AI’s accuracy rate on appointment scheduling? (Ask for data.)

Compliance and Security

12. How do you handle TCPA compliance? The FCC has explicitly stated AI-generated voice calls fall under the same restrictions as prerecorded messages.
13. How do you handle call recording consent in two-party states?
14. Where is call data stored? Who owns it?

Business Terms

15. What’s the contract term? Month-to-month, annual, or multi-year?
16. Is there an early termination fee?
17. Is pricing per-rooftop flat fee, per-call, or tiered?
18. What’s included versus what costs extra?
19. Do you have OEM co-op reimbursement eligibility?

Performance

20. How do I know if the AI is actually performing? (If the answer is “our dashboard shows calls handled,” that’s not enough.)

The AI Search Angle: How Dealers Are Finding These Providers

Most dealers first encounter AI phone agents at the NADA Show. All eight providers exhibited at NADA 2026. Second source: peer referral through 20-groups and DealerRefresh. Third: Car Dealership Guy’s newsletter reaches 55,000+ dealers, and Yossi Levi is a strategic investor in Mia Labs (and has featured Toma prominently), which means those providers get disproportionate exposure.

Google search interest for “AI phone agent” is still relatively small. But AI search is growing fast. LLM referral traffic to dealer sites is up 15x year-over-year (Fullpath). 9% of dealers are completely invisible to AI search platforms (Pied Piper 2026). If your customers are asking ChatGPT “best dealership near me” and getting your competitor, answering your phone faster doesn’t help if nobody’s calling. Our GEO guide covers the technical setup that gets your store into AI answers. And our guide to AI tools that actually work for dealerships covers how to evaluate which of these investments moves the needle versus which ones just demo well.

The agentic AI breakdown covers how these tools fit into the broader dealership AI stack, and our after-hours lead response guide covers the speed-to-lead side of the equation.

The BDC-versus-technology debate also matters here. Our BDC comparison breaks down when to invest in people versus systems.

The Bottom Line

Eight providers, $580 million in funding, real technology solving the missed-call problem. For service scheduling the ROI is clear. For sales, the math depends on what happens after the phone gets answered.

But people buy cars from people. Hybrid AI+human outperforms pure AI by 15-25% (Calldrip). Satisfaction drops 9 points at the handoff when the human doesn’t deliver (Pied Piper). AI answers the phone. A salesperson sells the car.

That’s what Ringlead does. Every internet lead gets a call back from your sales team in under 60 seconds. Every call recorded, including personal phones. Every conversation scored A-F so your managers know who’s following the process that actually sells cars.

Want to see what AI call analysis looks like when a real lead turns into a real phone call? Try the live demo and watch the handoff, recording, scoring, and manager alert flow.

Try the Live Demo

Frequently Asked Questions