DriveX Technologies

Meet Joseph—he discovered a chip on his windshield and contacted his local repair shop. Instead of going to the shop right away, he used a DriveX link to inspect the damage himself.🤳 In just a few minutes, with the help of AI, he completed the inspection, and AI instantly determined whether the windshield needed repair or replacement. In this video, we’ll walk you through the entire DriveX process for windshield jobs. Let’s take a look👇

New market unlocked - hey, Hungary! We’d like to introduce Ominimo Insurance, an insurtech that really does things differently. As you may know, most responsible drivers are overpaying for insurance, and Ominimo aims to change that. Through SIGNAL IDUNA Biztosító Zrt., they offer AI-driven third-party motor insurance and, as of this week, a smart casco solution. 🎉 Ominimo’s goal is to deliver the best possible prices for motor insurance, intelligently. Ominimo has integrated DriveX into their casco solution, allowing clients to photograph their cars in just minutes – aligning perfectly with Ominimo’s digital-only approach. We’re glad to be working with you, Gábor & the team!

2024 wrapped, but make it DriveX🚗

We have some exciting numbers to share🤩 In total, 50 000 clients have left feedback on our vehicle inspection web app SmartScan this year. 65%(!) of them have rated our solution 5 stars out of 5. …and 20% of them have given the rating 4. Needless to say we’re more than happy with these results. Of course, we’re always improving, but it’s motivating to see our solution making things easier for our customers. If they’re happy, we’re happy.😉

Did you know that Kentti’s all-time favourite artist is Justin Bieber? But that’s beside the point here, to be fair. We actually wanted to give you a backstage sneak peek into how we developed our windshield repair/replace decision AI. Kentti (our CTO, by the way) gives an in-depth overview of this in our latest blog post. The process was divided into five phases: 1️⃣ Planning: understanding the problem 2️⃣ Model selection: choosing AI model type and architecture 3️⃣ Data collection: gathering training data 4️⃣ Model training 5️⃣ Performance measurement: testing predictions against the test dataset Kentti pointed out that the most challenging part of developing this AI was gathering high-quality training data. To see how we solved this through a combination of manual effort and synthetic data generation, check out the blog post (link in the comments ⬇️).

Last week our CEO, Rauno Sigur, did a keynote at Äripäev’s conference for AI practitioners🎤 Rauno’s main points for companies building their own AIs were: ➡️ Avoid creating yet another ChatGPT wrapper—it’s easy to replicate, offers little control over outcomes, and lacks substantial intellectual property. ➡️ Genuine deep tech innovation typically requires 5-7+ years to reach market maturity. Are you prepared for that level of commitment? ➡️ Focus on dominating a small niche with global impact. How can you become the best in the world at something highly specific? The conference aimed to explore the transformative potential of AI through practical case studies, and we were more than happy to share our story. Thanks for having us!

It’s clear that repairing a windshield is cheaper and uses fewer resources than replacing one. But by how much exactly?👇 We’ve pulled together some insights to highlight the differences between repairs and replacements. For drivers with comprehensive car insurance, there's often a choice between the two, even when a repair would be fully sufficient. With DriveX windshield inspection technology, we support repair shops and insurers (and their clients of course) to choose repair whenever possible. Data sources: AXA, LKF, Carglass Estonia, and PZU.

TPAs (Third Party Administrators) for claims management offer insurers an effective way to lighten their workload and manage higher claim volumes. However, using TPAs does come with some controversies. Common objections include that: ➡️ The experience may feel less personal ➡️ Claims can take longer to resolve ➡️ There may be a higher risk of errors compared to in-house handling On the one hand, TPAs provide scalability and specialised expertise, which is essential during high-demand periods. On the other, ensuring a seamless and empathetic claims experience can be challenging when the process is outsourced. A digital TPA model might bring the best of both worlds. As Y Combinator noted in a recent post on ClaimSorted, some claims processes can be automated using AI, while human handlers focus on complex cases requiring personal attention. What do you think—are TPAs an efficient solution for managing growing claim volumes?

So, here’s where we’re at with AI in the insurance world — most insurances are still in the early stages, trying to figure it all out. There’s huge potential though, especially in core functions like customer service, underwriting, and claims. But... there are definitely some bumps in the road.   The biggest challenges? It’s not the tech itself but more about the people and processes — things like AI literacy and figuring out where to focus first. We’ve got the tools, but now it’s about making the pieces fit. And we’re here for it 😎 We're always ready to help insurers tackle their AI challenges, especially when it comes to vehicle inspection. Thanks for the great report, Boston Consulting Group (BCG)! 

We hit a big milestone last week – 1 million photos analysed live by our AI! 🥳 This means that, in total, DriveX software has analysed 1 million photos in real time from our clients. As the saying goes, AI is only as good as the data it’s trained with—and it needs plenty of it. Our engineers are working tirelessly to ensure that each image contributes to new and better models. Learning from real-world experience is in our DNA. One million is definitely a beautiful number, but fear not—it's only the start 🚀