Pricing
What it costs.
You pay for each robot running the agent. Seats are unlimited on every plan, so adding an engineer never changes the bill, and neither does adding a site or a project.
Free
$0
- 3 robots
- 25 captures a month
- 7 days of retention
- The full replay viewer, not a trial version
Team
Most teams$40 to $80
per robot, per month
Where you land in that range depends on fleet size and capture volume. Tell us your fleet size and we will give you the number.
- Unlimited seats
- 90 days of retention
- Slack, PagerDuty, Jira, and Linear
- SSO
Enterprise
Custom
Everything in Team, plus:
- Retention set by incident severity
- Private cloud
- Audit exports and legal hold
- A written response time
What counts as a robot
Any machine that has uploaded at least one capture in the billing month. A robot that sat in a maintenance bay all month is free.
Captures beyond the plan
Team includes a monthly capture allowance per robot. Past that you can either let extra captures bill per capture or have the agent stop uploading. You choose, and you can change it later. Both are shown on a meter in the dashboard before anything bills, and you can set a hard cap that we will not cross.
What we never bill for
Seats, gigabytes stored, or users viewing a replay.
Questions engineers ask first.
Why not just use a log viewer?
A viewer opens data you already have on your desk. Getting that data off a robot in a warehouse two states away is the part that takes six days. The agent is running before the failure happens, so the sixty seconds around it already exist and are already uploaded when you go looking. Once they are here, we group captures from different robots that share a root cause into one issue, which a viewer does not do.
Is the replay deterministic?
Playback is what ships today, and playback is not deterministic in any useful sense. It renders the recording. Re-running a single program against its recorded inputs is next, and that is deterministic for programs driven by messages rather than by hardware timing. Reproducing a full run bit for bit is not possible on a standard robot stack. The architecture page explains why.
What if our stack is not ROS 2?
Then the agent will not help you yet. It reads ROS 2 topics over Cyclone DDS or Fast DDS on Linux, arm64 or x86_64. A ROS 1 Noetic bridge and Zenoh transport are on the roadmap. If you are on something else, tell us what, because that is useful for us to know even if we cannot help today.
Start with three robots.
Free, no card, no call. Install the agent on three machines and see what it catches in a week.