Ottometric turns petabytes of raw sensor data into the handful of numbers that decide whether an automated-driving feature is safe enough to ship.
Above: the logo of a company whose whole job is to read the data nobody else wants to read - measured, like its name, in distance traveled.
Somewhere on a test track this morning, a vehicle drove itself for eight hours and recorded everything: cameras, radar, lidar, the lot. By lunch it had generated more data than a small city's worth of security footage. None of it matters until a human can answer one question - did the car do the right thing, every time, and can you prove it? That question is Ottometric's entire reason to exist.
The company sits in an office park on Winter Street in Waltham, Massachusetts, a short drive from the labs where a good chunk of American autonomous-driving talent was trained. Its software does something deeply unglamorous and quietly essential: it takes the firehose of data coming off Advanced Driver Assistance Systems - ADAS, the lane-keeping, auto-braking, adaptive-cruise machinery now standard in new cars - and tells engineers, in days rather than months, whether the system is actually working.
It is not a flashy pitch. There is no robotaxi in the lobby. But every company chasing autonomy runs straight into the same wall, and Ottometric built itself to live at exactly that wall.
Here is the open secret of the self-driving industry: gathering data was never the hard part. Bolt enough sensors onto a fleet and you will drown in petabytes by Friday. The hard part is the part nobody puts in the keynote - sitting with that mountain of recordings and turning it into a defensible verdict.
Traditionally, that work was manual. Teams of engineers scrubbed through footage, hand-labeled scenarios, argued over spreadsheets, and built one-off scripts to compute whether the system hit its key performance indicators. It was slow, expensive, and - the part regulators care about - hard to reproduce. When a validation run took months and cost millions, every design change meant starting the grind over.
Meanwhile the stakes kept rising. Europe's General Safety Regulation now mandates safety features that have to be proven. Euro NCAP's 2026 protocols raise the bar again. Every OEM and Tier-1 supplier suddenly needed to validate more systems, more rigorously, on a shorter clock - using a process that was held together by manual labor and good intentions.
Ottometric was founded in 2019 by Joseph Burke and a team of automotive veterans - people who had spent the previous decade inside General Motors, NVIDIA, Autoliv, and Optimus Ride, watching ADAS go from novelty to regulation. They had personally suffered the manual validation process. That is a useful kind of founder: the one who is genuinely annoyed.
Their bet was specific. Not another simulation tool, not another dashboard. Instead, a layer of AI that could do the distillation itself - reading raw, multimodal sensor streams and converting them into structured, KPI-ready insight automatically. If you could automate the judgment, not just the storage, you could collapse a months-long process into days.
Investors with automotive in their blood took the bet. The 2023 seed round, $4.9 million led by Rally Ventures, pulled in mobility specialists like Goodyear Ventures, Proeza Ventures, and Trucks VC - the kind of cap table that signals the people closest to the industry believed the problem was real.
A short history of refusing to read spreadsheets
Joseph Burke and a crew of GM, NVIDIA, Autoliv and Optimus Ride veterans set out to automate ADAS validation.
Rally Ventures leads, with Goodyear Ventures, Proeza Ventures, Automotive Ventures and Trucks VC aboard.
Tier-1 suppliers report validation cost cut in half and turnaround sped up 6x on real programs.
Schooner Capital leads; Rally and Proeza follow on, joined by PS27 and Somersault. Orhan Gazelle joins the board.
Euro NCAP 2026 and GSR mandates land - exactly the clock Ottometric was built to beat.
The platform is modular by design, because no two validation programs look alike. It breaks into four parts that stack into an end-to-end pipeline.
OttoViz is the part engineers fall for. Safety reviews used to dissolve into arguments over screenshots; OttoViz gives everyone the same interactive replay, with full traceability back to the source data, so a gate review becomes something you can actually defend rather than survive.
Under the surface sits a thoroughly modern AI stack - PyTorch and TensorFlow for the models, Spark and Apache Iceberg for the data, and vector databases like Pinecone, Milvus and Weaviate for retrieval. The buzzwords are real here; they are doing actual work.
Claims are cheap in automotive AI. Ottometric leans on results from real suppliers instead. A North American Tier-1 cut its validation cost in half and turned runs around six times faster. An EU Tier-1 raised annual profit on its Intelligent Speed Assistance product by 174% and compressed homologation - the formal sign-off that a vehicle meets regulation - from weeks to days.
Relative time-to-result on a Tier-1 ADAS program // bigger bar = slower
Source: Ottometric customer results (North American Tier-1). Figures are company-reported and approximate; treat them as direction, not audited fact.
Behind the customers sits a cap table that knows the territory: Schooner Capital led the 2025 Series A, with Rally Ventures and Proeza Ventures following their earlier checks, joined by PS27 and Somersault Ventures. Schooner's Orhan Gazelle took a board seat - the kind of commitment that comes with conviction, not just a wire transfer.
Strip away the jargon and Ottometric's mission is plain: shorten time to market and reduce total program cost for ADAS and autonomous-vehicle development. Three words sum up the pitch the company puts in front of customers - faster, cheaper, confident.
That last word matters most. The industry does not lack ambition or data. What it lacks is a fast, repeatable way to be sure - to stand in front of a regulator, or an internal safety board, and show the work. Ottometric is selling certainty at speed, which in a field where the downside is measured in human safety is not a small thing to sell.
It is a B2B story, sold to Tier-1 suppliers and OEMs, with roughly 48 people and an estimated $3.5M in revenue at last look. Small, focused, and aimed squarely at a bottleneck that grows more expensive every year the regulations tighten.
The self-driving future will not arrive because someone built a smarter car. The cars are already smart. It will arrive when the industry can prove, quickly and repeatedly, that the smart cars are safe enough to sell - across thousands of edge cases, under tightening regulation, on budgets that cannot absorb a months-long validation cycle per feature.
That is the wall Ottometric chose to live at. As GSR and Euro NCAP 2026 turn validation from best practice into legal requirement, the company's unglamorous specialty becomes a lot less optional. The fresh $10M is aimed precisely there: scaling the platform for a wave of suppliers who suddenly have no choice but to do this faster.
Return to that test vehicle from this morning - eight hours of driving, a small city of footage, one question hanging over all of it. In the old world, answering it took a team and a quarter. In Ottometric's world, the data grades itself overnight, points back to its own sources, and hands an engineer a verdict they can defend by breakfast. The car still has to drive. Now somebody can finally keep up with reading what it did.
Sources, socials & places to dig deeper