Adoption data is only as good as the agents collecting it.
A sync agent that stopped working looks exactly like a developer who never uses Claude Code. Your adoption numbers cannot tell them apart.
On day one of our own pilot, 3 of 22 installs were already broken, and those developers were flagged as low adoption for a full week.
Without fleet monitoring, nobody knows which machines run which agent version until something breaks.
Each installed agent sends a heartbeat with its health, version and last sync. Every machine shows as Online, Stale or Offline, so a silent failure is visible the same day instead of at renewal.
Every agent reports its version, so drift across the fleet is visible at a glance. Pin a version for the whole organization, roll new versions out in stages and roll back without touching a single laptop.
Machines that go Stale or Offline turn into alerts, measured against an uptime target for the team, and arrive where your team already works.
Numbers from our own 22-developer rollout.
What fleet monitoring gives engineering leadership, and what it proves to your auditors.
For CTOs and VPs of Engineering
For Security and Compliance
How heartbeats, alerts and version rollouts work.
It is a continuous health check of the agent installed on every developer machine. Each agent sends a heartbeat every 3 minutes with its health, version and last sync, so you know every install is working instead of assuming it is.
Install the agent once per laptop. Each one reports a heartbeat and its session activity, so the fleet view shows which machines are Online, Stale or Offline next to real usage, and low adoption is never confused with a broken install.
Within 10 minutes when an agent reports an error. A machine that stops sending heartbeats is marked Stale after 30 minutes and Offline after 24 hours, and both states raise an alert.
Hook crashes are isolated and never crash the session itself. The crash is logged locally and reported in the next heartbeat, so it shows up on the health dashboard.
Yes. Agent versions are pinned by default at the organization level. You can roll a new version out to part of the team first, then everyone, and roll back one organization without reinstalling anything on developer laptops.
Install the agent once per laptop and every machine starts reporting its health within minutes.
