Rerun
Rerun is a Stockholm-based developer tools company that builds an open-source data stack for so-called physical AI—robotics, autonomous vehicles, drones, and other systems that operate in the real world.
Profile
Rerun provides an open-source data platform for logging, visualizing, querying, and training on multi-modal sensor data from robots, drones, and autonomous vehicles.
Rerun is a Stockholm-based developer tools company that builds an open-source data stack for so-called physical AI—robotics, autonomous vehicles, drones, and other systems that operate in the real world. Founded in 2022, the company raised a $17 million seed round in March 2025, led by Point Nine Capital and with participation from Y Combinator, to expand its engineering team and bring a commercial product to market. Rerun's core offering is an open-source visualization and data management platform that lets engineers log, query, and inspect multi-rate, multimodal sensor data—lidar, camera feeds, telemetry, and neural network outputs—in a single viewer.
In early 2026, with the release of SDK version 0.32, the company rebranded its product as a unified data layer, adding a commercial backend called Rerun Hub that enables SQL queries across recordings, derived column transforms, and direct streaming to GPU training jobs. Rerun Hub is the company's first paid product, sold as a cloud or on-premises catalog and storage engine. The company has not disclosed revenue or headcount.
It maintains an active open-source community on GitHub and Discord, and its SDKs support Python, Rust, and C++. Rerun's investors include Point Nine Capital and Y Combinator. The company is led by co-founders Emil Ernerfeldt (CEO) and Andreas Reich (CTO), both former engineers at Embark Studios.
Rerun faces a crowded field of competitors including Foxglove, Weights & Biases, and Comet ML, and must convert open-source users into paying Hub customers to generate sustainable revenue. The company has not disclosed any named enterprise clients.
Who buys this
- Robotics startups and research labs building perception and control stacks
- Autonomous vehicle teams at OEMs and tier-1 suppliers
- Drone and UAV developers needing real-time sensor data inspection
- Computer vision and SLAM researchers debugging training pipelines
- Industrial automation engineers working with multi-rate sensor fusion
Strengths and what to watch
Strengths
- Open-source core with permissive license (MIT/Apache 2.0) has built a visible community on GitHub (over 10,000 stars) and Discord, lowering adoption friction for individual developers and small teams.
- Product architecture is designed for the specific data shapes of physical AI—multi-rate, multi-modal, time-series sensor data—rather than repurposing a general-purpose visualization tool.
- Commercial Hub product addresses a concrete pain point: streaming sensor data directly to GPU training without an intermediate export step, which could reduce pipeline complexity for robotics teams.
Watch for
- Revenue model risk: Rerun Hub is the only paid product; if conversion from free open-source users to paying Hub customers is low, the company may struggle to reach product-market fit in the commercial segment.
- Competitive pressure: Foxglove (also YC-backed, with a similar open-source visualization tool for robotics) and established ML platforms like Weights & Biases and Comet ML offer overlapping capabilities with larger sales teams and enterprise relationships.
- No disclosed enterprise clients or named deployments: the company's traction is primarily community-driven, and it has not publicly demonstrated adoption by large OEMs or autonomous-vehicle programs, which are the highest-value customers in this space.
Key Information
- Industry
- Computer Vision
- Founded
- 2022
- Headquarters
- Stockholm
Frequently Asked Questions
What is Rerun and what does it do?
Rerun is an open-source data platform for logging, visualizing, querying, and training on multi-modal sensor data from robots, drones, and autonomous vehicles. It helps engineers inspect lidar, camera feeds, telemetry, and neural network outputs in a single viewer.
How does Rerun handle sensor data for robotics?
Rerun is designed for physical AI data shapes—multi-rate, multi-modal, time-series sensor data. Engineers can log, query, and inspect lidar, camera feeds, telemetry, and neural network outputs in one viewer, simplifying debugging and training pipelines for robotics and autonomous systems.
What is Rerun Hub and what does it offer?
Rerun Hub is the company's first paid product, a commercial backend released in early 2026. It enables SQL queries across recordings, derived column transforms, and direct streaming to GPU training jobs, available as a cloud or on-premises catalog and storage engine.
Who are the main competitors to Rerun?
Rerun faces competition from Foxglove, Weights & Biases, and Comet ML. Foxglove offers a similar open-source visualization tool for robotics, while Weights & Biases and Comet ML provide ML platforms with overlapping capabilities and larger sales teams.
What programming languages does Rerun support?
Rerun provides SDKs for Python, Rust, and C++. These allow developers to integrate the platform into their existing workflows for logging and visualizing multi-modal sensor data from robots, drones, and autonomous vehicles.
How is Rerun funded and who leads the company?
Rerun raised a $17 million seed round in March 2025 led by Point Nine Capital with Y Combinator participation. The company was founded in 2022 and is led by co-founders Emil Ernerfeldt (CEO) and Andreas Reich (CTO), both former Embark Studios engineers.
Sources
- techcrunch.com — $17 million seed round led by Point Nine Capital with Y Combinator participation; company founded 2022; co-founders Emil Ernerfeldt and Andreas Reich; product positioning for physical AI.
- www.rerun.io — Product description, open-source SDKs (Python, Rust, C++), Rerun Hub commercial backend, use cases with LeRobot, Brush, PyCuVSLAM, and Project Aria.
- www.rerun.io — Announcement of SDK 0.32 and Rerun Hub as a unified data layer with SQL queries, derived columns, and GPU streaming.
- github.com — Open-source repository, license type, community activity.