NEWAsk your fleet questions in plain language. Meet Ask →
Ask · AI Troubleshooting

Explore your fleet. Understand what went wrong.

Ask questions about devices, running models and hardware. Investigate failed tasks, low model confidence and motor issues, all in one conversation.

Chat with your Physical AI fleetAnalyzeCompareMonitorDebug

Works with robots of every shape and size

  • Arms
  • Humanoids
  • Robot dogs
  • Mobile robots
  • Tractors
  • Delivery bots
  • Drones
  • Forklifts
  • Excavators
  • Cleaners
Explore

Explore every device and model.

See what each machine runs and how it performs. Narrow an investigation by device, model version, hardware or task.

Which robots have the lowest task success?
Scanning fleet · 14,203 records
6 results
DeviceRobot #1311
Running modelVLA v1.2.6
Task success75%
HardwareHybrid
Last update 24ms
Logs
On
Investigate

Follow the investigation.

Start from a failed task and ask what changed. Review confidence, sensor inputs, motor responses and recent updates, with suggested follow-up questions at every step.

Why is Robot-002 missing its target?
Analyze
Motor response delaysSeen in failed attempts
Likely cause
Model confidenceSimilar to successful runs
Normal
Next: Show the motor logs · Compare with similar robots · When did this start?
Evidence

See the evidence behind the answer.

Open the model traces, device logs and hardware signals behind each finding to see what happened and examine possible causes.

  • Model traces: inspect inputs and outputs
  • Device logs: software and system events
  • Hardware signals: sensor readings and motor responses
Evidence
Robot-002 · failed task
Model tracesInputs and outputs
Open
Device logsSoftware and system events
Open
Hardware signalsSensor readings, motor responses
Open
From fleet to motor

From a fleet-wide question to a specific machine, model or motor.

Ask in plain language

Ask which devices are struggling, what models they run, or what changed after an update.

Connect models and hardware

Relate model decisions to camera inputs, sensor readings, motor responses and software events.

Explore what changed

Focus on a failed task, a recent deployment or a time window to reconstruct events.

Inspect running models

Review model versions, confidence signals, inputs, outputs and inference latency on each device.

Compare across your fleet

Compare affected machines with similar devices on the same task to find meaningful differences.

Debug remotely

Inspect device logs, system state and model traces without being physically present.

FAQ

Questions, answered.

Something else? Talk to our team.

What can I ask?

Ask which devices are struggling, what models they run, or what changed after an update. Ask can narrow an investigation by device, model version, hardware or task.

Where do the answers come from?

Ask relates model decisions to camera inputs, sensor readings, motor responses and software events. You can open the model traces, device logs and hardware signals behind every finding.

Do I need to be on site to debug?

No. You can inspect device logs, system state and model traces remotely.

From your first machine. To the entire fleet.

See Physicalfleet run a live fleet, or talk to the team about yours.

Free for up to 5 devices · Works with robots of every shape and size