cloudparts

Foxglove

An observability platform for robots: a web and desktop app that replays and visualizes multimodal robot data, such as 3D scenes, camera video, plots and maps, on top of a data platform that stores, indexes and searches MCAP recordings.

What it is
Observability
Used for
Logs
Type
Commercial product
Made by
Foxglove Technologies Inc.
License
Proprietary; the Foxglove SDK (with Foxglove Bridge) and MCAP are MIT. Foxglove Studio 1.x, open source under MPL 2.0, stopped at v1.87.0 in February 2024
Facts checked
13 October 2026
Website

What it is

Robots log many parallel streams at once: cameras, lidar, IMU, plus the output of perception, planning and control. The Foxglove app, in a browser or as a desktop app, lays those streams out in panels (more than 20, including 3D, image, plot, map and state transitions) arranged into layouts a team can share. The same panels work on local files (MCAP, ROS 1 bags, ROS 2 .db3 files, PX4 ULog) and on live data: Foxglove Bridge streams from a ROS robot, and the open-source Foxglove SDK, a Rust core with Python and C++ bindings, streams from your own code over a local WebSocket while writing the same messages to an MCAP file.

Behind the app sits the Data Platform. Foxlet, which runs on the robot, uploads recordings, which are indexed by device, time and topic. Data Search queries messages, events, devices, recordings and sessions with no separate warehouse, and Events and Sessions mark and group the runs that matter. The data can live in Foxglove’s cloud, in your own S3, Google Cloud Storage or Azure Blob bucket (Bring Your Own Storage, indexed in place with read-only access) or in a self-hosted deployment. Remote Access, generally available since August 2026, streams a field robot’s topics over WebRTC through a Foxglove-managed selective forwarding unit, with no VPN, and carries teleoperation commands back.

Pricing is a free tier, Pro from $20 a month with usage billing, and custom Enterprise plans; students and researchers with a .edu or .ac address get it free. Foxglove started as an open-core project, but with Foxglove 2.0 in March 2024 the company merged its visualization app and data platform into one closed-source product and stopped developing the open-source Foxglove Studio.

Why teams pick it

  • One tool for recorded and live data: the same panels and layouts open MCAP files, ROS bags and live robots, with more than 20 built-in panels for 3D, images, plots and maps
  • Recordings stay in open formats (MCAP, ROS 1 bags), and Bring Your Own Storage indexes them where they already sit in S3, Google Cloud Storage or Azure Blob, without making a copy
  • Remote Access streams live topics from robots behind firewalls or on cellular links over WebRTC, and the selective forwarding unit gives each viewer its own bitrate
  • Free for individuals and small teams (10 GB, 3 users, 5 devices); Pro starts at $20 a month with 1 TB and 3 Developer Seats, plus unlimited free Basic Seats for people who only view

Watch out for

  • Closed source since 2.0: the last open-source Studio release was v1.87.0 in February 2024, so teams that need the source must maintain a community fork of that MPL 2.0 code themselves
  • Usage billing on several meters (storage, query time, indexing and egress bandwidth) plus $42 a month per extra Developer Seat and $20 per active device, so the bill follows usage; Foxglove offers a cost calculator to estimate it
  • Key features are tiered: Remote Access needs Pro or Enterprise, self-hosted data needs Enterprise, and the Grafana integration is an Enterprise add-on
  • Built for robot recordings, not service telemetry: uptime, deploy and cluster dashboards stay in tools like Grafana, and the agent sidebar and Semantic Search launched in beta in August 2026

Alternatives

What all observability have in common →Compare side by side →
Amazon CloudWatchAWSAWS’s built-in monitoring service: metrics from every AWS service, logs, alarms and dashboards, and now traces, OpenTelemetry and PromQL, billed per metric, per GB and per alarm.ApitallyCommercialAPI monitoring and analytics added with a few lines of middleware: per-endpoint and per-consumer traffic, errors and latency, plus optional request logs and traces, for Python, Node.js, Go and .NET web frameworks.Better StackCommercialUptime checks, on-call paging and status pages, plus logs, traces, metrics and errors in a ClickHouse-based store, sold as one SaaS billed per responder and per GB.ClickStackOpen sourceClickHouse’s open-source observability stack: an OpenTelemetry collector, ClickHouse for storage and the HyperDX UI for search, dashboards and alerts on logs, traces, metrics and session replays.DatadogCommercialA hosted monitoring platform: an agent on every host ships metrics, logs and traces to Datadog’s SaaS, which bills per host, per GB and per metric.GrafanaOpen sourceDashboards and alerts over data that stays where it is: Prometheus, SQL databases, cloud monitoring and more.IceGateExperimentalOpen sourceA Rust engine that takes OpenTelemetry data and keeps it in Apache Iceberg tables on S3, with Loki- and Tempo-compatible query APIs and Arrow Flight SQL.OpenTelemetryOpen sourceThe vendor-neutral standard for producing and shipping traces, metrics and logs, with SDKs for most languages and a Collector to route the data.ParseableOpen sourceA Rust server that stores logs, metrics and traces as Parquet files on object storage and queries them with SQL, without indexing at ingest.TreblleCommercialAn API platform that captures every request and response, payloads included, through SDKs or gateway plugins, and builds observability, analytics, documentation, governance and security checks on top of that traffic.

How it fits with other technologies

Log to an MCAP file and stream live to the Foxglove app with the SDK · python

# pip install foxglove-sdk
import math, time
import foxglove
from foxglove import Channel
from foxglove.channels import SceneUpdateChannel
from foxglove.messages import Color, CubePrimitive, SceneEntity, SceneUpdate, Vector3

scene_channel = SceneUpdateChannel("/scene")                 # a Foxglove message type
size_channel = Channel("/size", message_encoding="json")    # your own data

with foxglove.open_mcap("quickstart-python.mcap"):   # record to a file...
    server = foxglove.start_server()                  # ...and serve a live WebSocket
    while True:
        size = abs(math.sin(time.time())) + 1
        size_channel.log({"size": size})             # stamped with the current time
        scene_channel.log(SceneUpdate(entities=[SceneEntity(cubes=[
            CubePrimitive(size=Vector3(x=size, y=size, z=size),
                          color=Color(r=1.0, g=0, b=0, a=1.0)),
        ])]))
        time.sleep(0.033)

See it in a real stack

No architecture uses Foxglove yet. Drop it into the playground and wire it up yourself.Open playground →
Where it came from2021–2026

Adrian Macneil spent years leading infrastructure and developer productivity at the self-driving car company Cruise, where less than half of the engineers worked on the vehicle’s own software. The rest built CI/CD, fleet management, data infrastructure and simulation in-house, and colleagues at other autonomous vehicle and robotics companies were building nearly identical stacks. One of those tools was Webviz, Cruise’s open-source browser app for visualizing ROS data. Macneil and co-founder Roman started Foxglove to sell that kind of tooling off the shelf, beginning with a fork of Webviz released as a desktop app in 2021.

The company raised a $3.7 million seed round led by Amplify Partners (announced January 2022), launched the MCAP log format in February 2022 and raised a $15 million Series A led by Eclipse Ventures that October, by which time its user base had grown more than eight times in a year. It ran two products: Foxglove Studio for visualization, open source, and Foxglove Data Platform for data management, closed source.

In March 2024 Foxglove 2.0 merged the two into one product for web and desktop, introduced new pricing and ended the open-source edition of Studio. Since then the company has released the Foxglove SDK (April 2025), raised a $40 million Series B led by Bessemer Venture Partners (November 2025), cut Pro to $20 a month (April 2026) and, in August 2026, repositioned as an “agentic data platform for physical AI” with an in-app agent, an MCP server, Semantic Search and generally available Remote Access.

Why it’s built this way

A desktop app first
Leaving the browser sandbox let the first Foxglove connect directly to a ROS 1 stack and escape the browser’s performance limits on large amounts of robot data, while still running on Linux, Windows and macOS.
One product for visualization and data
Users said that separate visualization and data management products meant constant context switching and inconsistent web and desktop apps, so 2.0 merged the two codebases.
Close the visualization code
Visualization took most of the development effort and delivered most of the value to customers, while community contributions were under 1% of commits; the company said it couldn’t sustain full-time development while releasing most of its code as open source.
Keep the data format open
Foxglove argues that open formats are the only option for storing critical robotics data, so MCAP and the libraries around it stayed open source, and the MIT-licensed SDK writes MCAP files.

Timeline

  1. 2021

    Foxglove launches a desktop visualization app forked from Cruise’s Webviz

  2. 2022

    $3.7M seed, the MCAP file format and a $15M Series A

  3. 2024

    Foxglove 2.0 merges Studio and Data Platform; open-source Studio ends at v1.87.0

  4. 2025

    Foxglove SDK released; $40M Series B led by Bessemer Venture Partners

  5. 2026

    Pro from $20 a month; agentic data platform and Remote Access over WebRTC

Sources15 · facts checked 13 October 2026

Improve this page on GitHub