Car repairs are more accessible with AI

A Brazilian startup harnesses the power of AI to estimate how much your car repair will cost.

The Brazilian app Car10 helps customers get a quick estimate of how much a car repair could cost, thanks to the help of artificial intelligence (AI)

Car10 solves one of the biggest problems of people when they suffer a minor car accident: knowing how to repair their car.

Among the journeys that people must pass to repair a car is knowing who to call, the trouble of traveling places to get estimates, and the constant concern that whoever they work with do not end up taking advantage of customer ignorance.

The Brazilian startup has created an app that allows customers to have a better perspective of the damage their car suffered. With this app, users have to send a photo and get three to five estimates from nearby automotive workshops that have been previously evaluated by Car10 on issues of quality and trust. The startup even guarantees that it will perform a free repair if the customer is not satisfied with it.

“We remove the fear of the process, the concern that they will take advantage of you.”

José Tafner, CFO of Car10

Innovation enhanced with AI.

The company located in Sao Paulo uses artificial intelligence to accelerate the process. Now, users can almost immediately give the user an approximate idea of an estimated cost of repair thanks to the adoption of Microsoft’s Azure Cognitive Services Custom Vision Service.

With the current system, users who send a photo receive a quote in about 30 minutes and one hour. With the new AI tools, Tafner commented that they could get a general sense of how much the repair could cost in about 30 seconds.

“It goes back to customer needs. When you have an accident or small crash, what you want to know is how much it will cost,” said Tafner. “The first need is speed and a certain level of accuracy.”

The AI system uses a machine learning model (ML) to compare damage to the customer’s car with other examples of similar damage to generate a reasonably close estimate. Then, the company works with repair shops to get firmer deals.

The AI system may accelerate the quotation process. Still, it does not replace the practical involvement that Car10 has in ensuring that customers feel comfortable during the process of repairing their car.

Tafner commented that Car10 works with clients throughout, from providing the estimate to schedule the visit and even paying through the Car10 digital platform. Then, the client has the opportunity to rate their experience and the workshop where the repair was made.

“The digital part of this journey is small. The biggest part is analogous,” said Tafner.

Focus on quality

Car10 has about 100,000 clients and works with approximately 4,000 automotive workshops in Brazil, ranging from large companies to small family workshops. Tafner commented that initially, the company focused only on large workshops, because of the idea that this was what the client preferred. But they found that customers didn’t care if the workshop was in someone’s garage or an elegant office.

“They care about the quality of service,” he said.

Car10 started in 2014 with three brothers who had previously worked for their father’s insurance adjustment business. When that business was sold, they decided to use their experience in the car repair industry to immerse themselves in the world of startups. Tafner joined a few years later, after decades of global experience in the corporate world. The service is designed for people who pay for repairs on their own, instead of relying on car insurance.

From the beginning, the leadership team, made up of four people, has relied heavily on technology and data. They run the Microsoft Azure cloud service, use Power BI dashboards, and built the application on the .NET framework.

“We love data all four. We constantly use them to improve the business,” said Tafner.

Even so, Tafner commented that as with many businesses that swim in data, it could be challenging to discover what pieces of data are useful.

They found a clear winner: The photos of car repairs. 

Car10 was able to use that data to train the ML model to automatically detect what type of repair the person’s care needs and how much it could cost in broad strokes. Car10 does not sell customer data and protects personal information through Azure security protections.

Car10, which has received investment funds for startups from Microsoft, first began building the AI solution when the company participated in an industry hack fest. Although it has IT, staff, none of the people who work at Car10 have particular experience in AI. Azure Cognitive Services is designed so that it can be used even by people without formal AI training.

Medium-term expansion

Now, Car10 is around five years old and expects to reach a balance in the next quarter. Tafner said that now, the company is looking for more funds to expand to other business areas and, potentially, to other markets outside Brazil.

One of the advantages that AI offers to business is the possibility of scaling.

“What we can do for car crashes we can do for other things,” he said.

For Tafner, the small team and the fast pace are stimulating and enlightening. As in any startup, he says, the company tests new things quickly, makes mistakes, and adjusts – all while trying to run the core business. He compares it to driving a race car.

“We changed the tires while the car is running,” said Tafner. “For us, there are no pit stops.”