Nexar Inc.

101,000 work zone crashes a year. Over 40,000 injuries. Nearly 900 fatalities - that's the reality Nexar and IBM set out to change, and 3 days ago, we did just that. Our AI-powered work zone compliance agent is now live in the IBM watsonx Orchestrate Agent Catalog. For decades, transportation agencies have managed work zones with permits, plans, and hope. Manual inspections. Contractor self-reporting. Citizen complaints. No continuous visibility into what's actually happening on the road. In the last decade alone, work zone fatalities climbed over 50%, yet the oversight model hadn't changed. Now it has. 350K+ connected cameras detecting cones, barriers, lane closures, and signage as they appear. Continuously. From real-world observation. Not when someone files a report. Not when an inspector shows up. As it happens. Swipe through to see the visibility gap we're closing. Link to the release blog in the comments. #Nexar #IBM #watsonx #RoadIntelligence #SmartCities #WorkZones

Nexar and IBM just changed how transportation agencies see their road networks. For decades, agencies issued permits and hoped for the best. Manual inspections. Contractor self-reporting. Citizen complaints. That was the system. Today, Nexar's AI-powered work zone compliance agent is live in the IBM watsonx Orchestrate Agent Catalog. It detects work zones as they appear. Barrels, cones, barriers, lane closures. Maps them in real time. Flags potential compliance issues automatically. Not when someone asks. Continuously. That is what an agent does. Real-world road intelligence, delivered through IBM's enterprise AI platform. No custom integrations. No months-long implementation cycles. Roads were one of the last major infrastructure categories managed without real-time data. Nexar and IBM just fixed that. https://bit.ly/4sOJfJe #Nexar #IBM #watsonx #RoadIntelligence #SmartCitie

Everyone's racing to build the best AI model. But most are racing with a blindfold on. Physical AI models improve with data. More data means more edge cases, more behavioral diversity, more rare-event coverage. This is well understood. What's less discussed is what happens when the scale of your data network compounds over time. Each new Nexar deployment generates new data. New data improves the model. A better model makes the product more valuable. A more valuable product attracts more deployments. The loop closes and tightens with every iteration. This isn't a feature advantage. It's a structural one. You can't shortcut a data flywheel by standing up a test fleet - not when the flywheel has been running for years, across hundreds of thousands of real vehicles, on real roads, in real conditions. The moat in Physical AI isn't the algorithm. It's the reality the algorithm was trained on. And reality, at this scale, took a long time to collect. What do you think is the most underestimated competitive moat in AI right now? For more on our network, and how it's powering the next generation of models, visit us here: https://bit.ly/4rQRJ1b #PhysicalAI #DataFlywheel #AVData #RealWorldAI #CompetitiveAdvantage

Morgan Stanley just upgraded Lemonade to Overweight. Stock jumped ~16%. The reason: a single Tesla partnership that gives Lemonade access to real-world FSD driving data - and the ability to price AV risk with it. Morgan Stanley's analyst wrote that autonomous driving vehicles will likely reshape how auto insurance is underwritten over time, and that new winners will emerge as autonomous vehicles reset competitive advantages. Lemonade earned that rating with one OEM and one dataset. The signal for the insurance market is clear: the insurers watching this are now asking "how do we get access to this kind of data at scale?" Lemonade's Tesla partnership provides unique access to vehicle data, enabling real-time risk assessment and differentiation between autonomous and human driving. That's powerful - but it's a single make, single behavior set. Nexar's network spans 350,000+ cameras, 100M+ miles of real driving data captured every month, across 98% of U.S. roads - every make, every model, every condition. The data layer that wins AV risk pricing won't be the one tied to one OEM. It'll be the one that sees everything else. See our CEO Zach Greenberger discuss the importance of scale in the clip below, and visit us here for more: https://bit.ly/4rQRJ1b #PhysicalAI #AVInsurance #RoadIntelligence #Nexar #InsurTech

Meta's V-JEPA 2, is pushing AI toward something essential: Understanding the world the way people experience it. Not as clean frames. But as uncertainty, negotiation, surprise. For communities, this distinction is enormous. It determines whether an automated system behaves predictably near a school zone. Whether it hesitates correctly around a cyclist. And whether it adapts when the unexpected happens. World models require immersion in authentic reality. That is where Nexar contributes, by providing continuous exposure to how infrastructure and human behavior truly interact - the awkwardness, the ambiguity, the imperfect choreography of real streets. Physical intelligence matures when it is accountable to the environments it will serve. And those environments are human. For more on us, visit here today: https://bit.ly/4rQRJ1b #PhysicalAI #PublicTrust #Autonomy

The average FNOL process takes days. What if it didn't have to? First Notice of Loss is traditionally the beginning of a slow, contested process. An adjuster call. A police report. A driver statement that may or may not match the physical evidence. Days of back-and-forth before anyone has a clear picture of what actually happened. Using physical AI, we can compress that timeline to seconds. The moment of impact triggers an automatic compilation: crash video, telematics, impact vectors, pre-crash behavior, location data, and an ML-based severity score - all assembled before the adjuster picks up the phone. The result isn't just a faster claims process. It's a more accurate one. The physical record exists, and the dispute doesn't have room to grow - not by removing humans from the process, but by giving them everything they need before the first conversation starts. Fewer disputed claims, faster resolutions, and adjusters who spend their time on judgment calls - not chasing evidence. The physical world captured the truth. Physical AI delivers it. How do you see physical AI assisting the claims chain? Visite here today for more: https://bit.ly/4rQRJ1b #PhysicalAI #FNOL #FleetInsurance #ClaimsAutomation #FleetSafety

Before you deploy a test fleet in a new city, you need to know what you're going to encounter. Most AV companies find out the hard way. Operational Design Domain expansion is one of the most expensive steps in autonomous vehicle development. You need to understand regional driving patterns, local edge cases, infrastructure conditions, and the behavioral norms of a geography before you trust your model to operate there. Traditionally, that means sending vehicles, which means time and cost before you've turned a wheel in production. Physical AI changes the economics of this problem. If the data from that geography already exists - structured, outcome-verified, and searchable - you can understand a new ODD before your fleet ever arrives. Nexar covers 98% of US highways. Every state. Every major corridor. Every regional driving behavior that exists in the country. The intelligence is already there. The question is whether you use it before you deploy, or learn from the field after. What do you think is better? For more: https://www.nexar-ai.com/ #PhysicalAI #ODDExpansion #AVData #AutonomousVehicles #RealWorldAI

Selling into government is a different game. RFPs are buried across dozens of decentralized state and local portals - and by the time most teams find them, the window has closed. So our GTM ops lead built an agentic system to solve it. No prior technical experience. Just a problem worth fixing and the tools to fix it. This is how we think about agentic AI at Nexar - not as a novelty, but as leverage. Our CBO Jon Miller wrote about it below #AgenticAI #CityStream #PhysicalAI #GovernmentTech

Most autonomous vehicle models knows what a person looks like. Very few can predict what that person is about to do. Vulnerable road users - walkers, cyclists, e-scooter riders, children - are the hardest prediction problem in Physical AI. Not because they're hard to detect. Because their behavior is context-dependent in ways that no vehicle model can fully anticipate. A person at a crosswalk behaves differently than a person mid-block. A cyclist in a bike lane behaves differently than a cyclist merging into traffic. An adult looking at their phone behaves differently than a child who just heard their name called. No simulation can completely capture these, because in the real world each instance looks different from the last. The only way to completely capture them is to observe them across real environments - the kind of environment our network captures continuously across 98% of US highways. Get this wrong and it's not a benchmark that suffers. It's a person crossing the street. For more: https://www.nexar-ai.com/ #PhysicalAI #VRU #AVSafety #PedestrianSafety #AutonomousVehicles

Before you underwrite a commercial fleet, ask yourself - what do you actually know about where they drive? Not the routes they file, nor the safety program they describe in their application, but rather what their drivers actually do, in the actual conditions they operate in, every day. The pre-bind intelligence gap is one of the most underestimated risk factors in commercial fleet underwriting. Fleets that look identical on paper can have radically different risk profiles - by geography, by time of day, by driver behavior, by infrastructure exposure. Nexar's pre-bind risk intelligence closes that gap. Real-world incident data, road risk scores by corridor, and behavioral pattern analysis - before the policy is written, not after a claim is filed. Underwriting risk you can see is a different business than underwriting risk you're estimating. What information do you wish you had before binding a commercial fleet account? For more: https://lnkd.in/dAzhgYUw #FleetInsurance #CommercialAuto #RiskIntelligence #Underwriting #FleetSafety