Car dealers have long been plagued by problems such as missed calls and high customer service churn rates, limiting customer experience and business development. Toma goes deep into the front line of the industry, and customizes exclusive AI agents for each dealer with unique AI voice technology to achieve efficient phone handling and high-quality customer service in response to the pain points of dealers. Its AI system can deeply integrate dealer multi-system data and improve service capabilities through continuous learning.
Can you imagine a car dealership missing 45% of calls every day? This means that almost half of the customers who want to make an appointment for maintenance, ask about accessories or consult about car buying are left aside. In an industry with an average profit of only a single-digit percentage, this loss is simply catastrophic. What’s even more troublesome is that the turnover rate of dealerships’ telephone customer service departments (BDCs) is close to 50%, almost double the national average. As I delved into the question, I discovered a startup that was revolutionizing automotive retail: Toma. They just secured a $17 million Series A funding round led by the Anderson Horowitz Fund (a16z), and their solution is making more than 100 dealerships across the U.S. rethink the nature of customer service.
The beginning of this story is interesting. Founders Monik Pamecha and Anthony Krivonos did not originally plan to enter the automotive industry, and when they founded the company in early 2024, they focused on AI voice products in the banking and medical fields. But a phone call from an Oklahoma dealer changed everything: “We’re almost overwhelmed with calls, you guys have to come and take a look.” “In this way, two Silicon Valley engineers embarked on a journey to a car dealership in the middle of the United States, which not only allowed them to find real business opportunities, but also showed me how AI can create the most real value in the most traditional industries.
Deep into the front line: the road of car dealership training of Silicon Valley engineers
What I think strikes me the most is the rare entrepreneurial spirit that Toma’s founding team shows: instead of sitting behind closed doors in a Silicon Valley office, they really go into the world of their customers. When the Oklahoma dealer saw the two young men clearly bearing the label of Silicon Valley engineers, he said directly: “You guys look like typical Silicon Valley engineers who don’t understand what’s going on here.” I’ll make a plan for you. Over the next few weeks, Monik and Anthony were “thrown” into the seven-store dealer group and asked to talk to everyone, from parking lot managers to owners, interviewing a total of 400 people.
To achieve these three challenges, product managers will only continue to appreciate
Good product managers are very scarce, and product managers who understand users, business, and data are still in demand when they go out of the Internet. On the contrary, if you only do simple communication, inefficient execution, and shallow thinking, I am afraid that you will not be able to go through the torrent of the next 3-5 years.
View details >
What impressed me even more was that they didn’t just interview, they actually answered the phone in the BDC department and experienced the challenges of customer service firsthand. Monik later recalled that this kind of work “quickly became boring” because there were too many repetitive tasks. But it was this experience of going deep into the front line that allowed them to truly understand the nature of the problem: every employee wanted to provide excellent service to customers in the morning, but as the day progressed, the burden of repetitive work and the ever-increasing expectations of customers left employees feeling exhausted and frustrated. This is not a problem of employee attitude, but a systemic problem.
During their field research, they discovered a statistic that shocked me: an average of 45% of incoming calls were missed by car dealers across the United States. The meaning behind this number is huge. Imagine if you run a restaurant and nearly half of your customers choose another restaurant because they can’t get a seat, how much of an impact will that have on your business? And in the automotive industry, this happens every day. Distributors spend hundreds of thousands of dollars on marketing and operations to deliver a five-star customer experience, but they fail in the most basic phone communications.
Their solution is ingenious: start with the simplest. Anthony programmed directly in the BDC office, and they let the AI handle the first 10-15 seconds of the call to see how far it could go. If the situation is not right, immediately transfer to manual customer service. With this incremental approach, they gradually expand the range of conversations that AI can handle until they can complete a full customer interaction. This approach reminds me of a lot of successful technology products: instead of trying to solve everything in the first place, focus on one core pain point and do it to the extreme.
What I especially appreciated was their experience in Oklahoma and Mississippi. Seema Amble, partner at a16z, who led the round, put it to a funny description: “They actually live with these dealers, attend dealer family barbecue parties, and really understand how they operate. “This reminds me of the common trait of the founders of the most successful vertical AI companies: they all live with their customers and have a deep understanding of how the industry really works. Monik even said that his wife was surprised by the oil stains on his clothes when he came home, indicating that they really got their hands dirty in the workshop.
What’s more, this on-the-ground experience led them to a key insight: a car dealership looks like a building from the outside, but inside there are actually 25 different departments, like 25 small companies operating under one roof. Each department has its own operational details and special needs, from finance to accessories to services, with a level of complexity that far exceeds their expectations. And these dealers sell the most expensive items an ordinary person can buy besides a house, but the customer experience has been neglected for a long time. This discovery made them realize that this is not just a technical problem but a systemic challenge that requires a deep industry understanding to solve.
Technological breakthrough in AI voice assistants: not just sounding human
When I took a closer look at Toma’s technical implementation, I found that their innovation goes far beyond making AI “sound human.” Their real breakthrough is to make AI “work like the best employee”. Traditional phone systems have accustomed us to the frustrating experience: “press 1 to sales, press 2 to service…” But Toma’s AI can directly understand the customer’s natural language needs and has access to all relevant system data from the dealership to provide accurate service.
What I find most impressive is their ability to “personalize”. Unlike that one-size-fits-all solution, Toma creates unique AI agents for each reseller that are as unique as a fingerprint. This customization is not just about changing the accent or speed of speech, but also about deep business logic customization. For example: Does this dealer offer free oil change service for life? What are the special policies for repeat customers? Under what circumstances will a mobility scooter be provided? Do you always mention the price when talking about the cost of diagnosis? These seemingly nuanced differences are key factors in determining the customer experience.
The reinforcement learning techniques they use are particularly ingenious to me. In simple terms, AI will try to solve problems on its own, and it will seek human help when faced with boundaries or uncertain situations. The key is that AI continues to listen to and learn from the processing process of human customer service. So the next time it encounters the same problem, the AI knows how to respond. It’s like an intern who is always learning, except it learns much faster than a human and never forgets.
What’s more, Toma’s AI can access multiple systems of the dealership at the same time: DMS (Dealer Management System), scheduling system, CRM, etc., and can read all relevant information at the same time. When a customer calls, the AI already knows their car purchase history, recent repairs, the content of the last 10 calls, and more. This information integration ability is unmatched by human customer service, because humans can only look at one screen at a time, while AI does not have this limitation.
From the perspective of customer acceptance, I found an interesting trend of change. Monik told me that when he first started, about one in five customers immediately refused to communicate when he heard that it was AI, but now this percentage has dropped to 4%. This change reflects a broader cultural shift: people are becoming accustomed to interacting with AI in hospitals, CVS pharmacies, hotels, and more, creating a better environment for car dealerships to adopt AI voice assistants. As Monik says, it’s like the transition from a human operator to a phone voice menu back then, a natural evolutionary process.
Practical results: Martin Management Group’s success story
No matter how good the theory is, it ultimately depends on the actual effect. Martin Management Group is a group that operates 16 dealerships nationwide and has been named a top 150 dealer group by Automotive News. The challenge they face is typical: increasing calls but not being able to increase BDC staff indefinitely to cope with peak periods, as recruitment, training, and payroll costs are rising, while also ensuring consistency in service quality.
After deploying Toma’s AI agents, Martin Management Group achieved impressive results in the first 90 days: scheduling more than 9,000 service appointments (generating more than $2 million in revenue), automating more than 22,000 calls, and reducing BDC workload by more than 40%. What’s more, this improvement doesn’t sacrifice customer experience, but frees up human agents to focus on more complex customer needs and relationship building.
I especially like the “retrieval” safety net mechanism they designed. When the AI forwards the call to a human agent, but the human agent does not answer in time, Toma automatically takes over the call again, leaves a message for the customer, and creates a detailed follow-up ticket in the task tracking software. This design avoids customers being forgotten during the transfer process, ensuring that every customer receives a timely response. This attention to detail reflects product design thinking that truly understands business needs.
Chadwick Martin, president of Martin Management Group, speaks eloquently about the results: “We were always looking for ways to improve BDC efficiency without adding staff or impacting the customer experience, and Toma did it. It took on repetitive work, handling tens of thousands of calls from our group, giving our team time to focus on higher-value interactions. It’s a game-changer. “This feedback from front-line managers is more convincing than any technical indicator.
From an operational point of view, I think the most valuable thing is the reallocation of time. When AI took on 50% of the calls, the team at Martin Management Group began to use the time saved for proactive outbound calls, including contacting lost leads, conducting recall campaigns, and other direct revenue-generating activities. This shift from reactive response to proactive attack is where the real value of AI lies: not simply replacing humans, but giving them time to do more valuable work.
One specific case that impressed me the most was a specific case: there was a dealer group with a centralized BDC team of 40 people, and after using Toma, AI took on about 50% of the incoming calls. What did they do with the time they saved? Six times more active outbound calls were made than in previous history. This means they can reach out to potential customers who have fallen silent, conduct recall campaigns, and follow up on opportunities that generate direct revenue. I find this 6x figure particularly convincing because it shows not a small increase in efficiency, but a fundamental change in the operating model.
What makes me think more is the “resolution rate” indicator of AI. Toma was able to achieve a 75% complete resolution rate, which means that three-quarters of customer interactions resulted in a satisfactory resolution without human intervention at all. This number reflects not only technical prowess, but also the depth of business understanding. AI knows how to view a customer’s purchase history, recent repair history, past calls, and even recommend relevant recall repairs. As Monik puts it, this AI “has read manuals for all possible models, has all the pricing information, can do mathematical calculations, and even has a PhD in physics”, but the key is that it combines these capabilities with the customer’s specific situation to create personalized experiences that have never been possible before.
Behind the financing: why a16z is betting on car dealer AI
When I saw a16z leading Toma’s $17 million Series A funding, my first reaction was: Why would a top Silicon Valley VC be so interested in what seems like a traditional auto dealership industry? After digging deeper, I found that there is a very clear logic behind this.
The scale of the U.S. auto industry is staggering: there are about 18,000 licensed new car dealers nationwide, with total sales of more than $1.2 trillion in 2024, more than 270 million repair orders completed, and more than $156 billion in service and accessories revenue alone. This is not a niche market, but an important pillar of the US economy. More importantly, the industry is relatively low in digitalization, leaving huge space for the application of AI technology.
Explaining the investment logic, Seema Amble, a partner at a16z, said, “We have invested in a lot of next-generation vertical AI companies, and the best founders have lived with these clients and really understand their operations. This sentence points to the core of vertical AI investment: not how advanced the technology is, but the ability to solve real problems with a deep understanding of a specific industry. Toma’s founders’ field research in Oklahoma and Mississippi exemplifies this deep understanding.
I think another key factor that a16z is interested in is the maturity of the market timing. AI voice technology has reached a level that can be actually deployed, and customer acceptance of AI interaction is also increasing rapidly, while the cost pressure and turnover issues faced by dealers have created a strong demand for AI solutions. This combination of technology maturity, market demand, and customer readiness is a key element of a successful investment.
From a business model perspective, the subscription model adopted by Toma is also attractive. As AI agents can handle more parts of a dealership’s operations, resellers will need to pay for these additional capabilities. This value-based pricing model is a sustainable business model that guarantees both ongoing revenue and is linked to actual customer revenue. Monik mentioned that they have a spreadsheet with over 300 functional requirements, which are unsolicited by dealers, indicating that the market demand is far from saturated.
What makes me even more optimistic about the team’s execution is its performance. They only hired their first real sales employee in recent weeks, and the previous growth was entirely based on product word-of-mouth spread and the founder’s personal promotion. This growth method driven by product strength is very rare in the field of enterprise services, which shows that they have indeed solved the real pain points of customers.
Deep Thinking on Industry Change: From Tools to Partners
I think Toma’s success story shows us an important trend in the application of AI in traditional industries: from providing tools to becoming partners. Traditional software is tools that users need to learn how to use; AI agents, on the other hand, are more like partners, understanding the user’s needs and actively helping them achieve their goals. This shift is particularly significant for industries like car dealerships, where turnover is frequent and training costs are high.
In my conversations with distributors, I found that the real challenge they face is not technical issues, but how to deliver a consistent, high-quality customer experience in a highly competitive and thin margin environment. Car dealerships primarily derive profits from services rather than sales, and customer retention is their holy grail. Every missed call means lost revenue and lost customers, which is unbearable for distributors with only a single-digit percentage of net profit.
Toma’s AI agent solves not just the problem of answering calls but the consistency of the entire customer experience. When AI can remember all of a customer’s historical information, understand their preferences, and can provide personalized service at any time, the improvement of customer experience is a qualitative leap. As Monik describes his vision for the future in the podcast: “Customers will be pleasantly surprised by the experience, both in-store and out-of-store experiences will improve dramatically. The salesperson loads all the background information ahead of time and knows what you like and what you don’t like, like a friend selling you a car, not a stranger. ”
One thing I particularly agree with is that good AI should surprise customers rather than frustrate them, and allow employees to save time rather than increase workload. This is the key criterion that distinguishes excellent AI products from ordinary AI products. Toma does a great job of this, and they even have an internal metric to measure how “surprised” customers are, analyzing tone and word choice to judge whether customers are surprised by the AI’s performance.
More broadly, I think Toma represents the right direction for AI applications in vertical industries: a deep understanding of industry-specific needs, customized solutions, and the ability to integrate seamlessly with existing systems. This approach, while seemingly limiting the market size, can actually create greater value because it addresses real and urgent business needs.
My prediction of the future of AI voice assistants
Based on Toma’s success in the car dealership sector, I have several bold predictions about the future development of AI voice assistants. I think we are at an inflection point: AI will no longer be a dispensable aid, but will become a core part of business operations.
In customer service, I predict that traditional “push-to-to-forward” phone systems will largely disappear in the next 2-3 years, replaced by AI agents that can understand natural language and solve problems directly. This shift will not only improve efficiency but also change customer expectations for service quality. When people get used to having smooth and natural conversations with AI, it is very inconvenient to return to mechanical button selection.
I am particularly optimistic about the potential of AI agents in knowledge-intensive industries. Car dealerships are just the beginning, and industries such as healthcare, legal, finance, and others that need to deal with large amounts of specialized information and complex processes face similar challenges. The advantage of AI is that it can grasp all relevant information at the same time, does not affect the quality of service due to fatigue or emotions, and can work 24 hours a day.
But I also see some challenges to be aware of. Privacy and data security will become increasingly important issues, especially when AI needs access to detailed historical information about customers. Businesses need to find a balance between providing personalized service and protecting customer privacy. In addition, the transparency of AI decision-making is also an issue that needs to be addressed, and customers have the right to know how AI makes recommendations or decisions.
From an HR perspective, I don’t think AI agents will simply replace humans, but will redefine job roles. Cases like Martin Management Group show that when AI takes on repetitive tasks, humans can focus on building higher-value customer relationships and solving complex problems. This requires companies to rethink personnel training and career development paths, ensuring that employees can collaborate with AI and leverage their strengths.
Most importantly, I believe that a successful AI voice assistant must possess three core qualities: being technologically advanced enough to provide a smooth interactive experience, having a deep understanding of specific industry needs, and being able to integrate seamlessly with existing systems operationally. Toma excels in all three areas, which is why they are able to stand out in the highly competitive AI market.
Looking ahead, I think there will be more and more vertical AI companies like Toma, and dedicated AI solution providers will emerge in every industry. This trend of specialization will drive the in-depth application of AI technology in various verticals, ultimately changing all aspects of our work and life. The intelligence of car dealerships is just the beginning, and bigger changes are yet to come
