Use Cases
Fleet Monitoring
Catch fleet problems before the field reports them. Alloy checks every run against your own checks, flags the robot that needs a person, and rolls up what keeps recurring.
What it is
Checking every run from every robot against your team's own checks, so the robot that needs a person is found before the field reports it.
How Alloy does it
- Every run gets a review. A recurring report runs your checks on each new run after it uploads. New organisations start with a First-look report on every run.
- Only the runs that need a person reach the team. A second report writes only when your criteria fail, and posts to the Slack channel you choose.
- Patterns roll up. A daily or weekly report counts what recurred, on which robots, with links to the runs.
- A pattern becomes a check. Ask the agent to turn it into a scenario, and every new run is checked for it.
Ask in the Alloy app, in Slack, on a Jira ticket, or from your coding agent.
Example
Fictional example: the robots, routes and numbers are invented.
A team runs 40 sidewalk delivery robots across three suburbs. They set up three recurring reports.
A. Review every run (Slack off): "Review this run. Report deliveries completed, stops longer than 30 seconds and why, wheel-slip events, and battery at return. Rate the run 1 to 5 against our criteria." It runs on each new mission and writes a review on every run. It posts nothing to Slack.
B. Flag the runs that need a person (posts to #fleet-alerts): "Write a report only if this run has more than 5 wheel-slip events, a stop longer than 60 seconds, or battery at return below 25%. Otherwise write no report." It runs on each new mission. Slack gets a message for each report it writes, so the channel hears only about the runs that fail the criteria.
C. Every Monday: "Across last week's runs: runs per robot, the five most common causes of long stops with counts and example links, and anything that got worse since the week before."
On Wednesday night it rains. Overnight, 212 runs upload. Each gets a review from A. Automation B writes seven reports, and #fleet-alerts gets seven messages:
DR-118, Hillcrest loop. 14 wheel-slip events on the Quarry Road ramp, against 0 to 2 on dry days. Two stops longer than 30 seconds, both on the ramp, both cleared by the robot. [Run]
By Thursday morning, six of the seven point to the same ramp, on five different robots. The operations lead asks the agent to watch for wheel slip on that ramp. The agent creates a scenario. Each new run from any robot on that route is now checked for it, and the past week of runs is searched too. On Monday the weekly report shows the ramp as the top cause of long stops, with every run linked.
Nobody opened 212 runs. The two engineers who look at the ramp start with the evidence. Deciding what to change on the robot or the route is still their call.
How to start
Ready to start. Talk to us. Alloy sets up your workspace with you, connects your fleet's runs, and writes the per-run review and alerts from what your team checks today.
Want proof first. Connect a few robots. Every new run gets a first review as it uploads, and you see the first alerts and the weekly roll-up before adding the rest.
Just looking. See the agent work on sample robots in Try Alloy.
What we need from you: access to your bucket or Alloy Edge on the robots, the Slack channel for alerts, and the checks your team runs today.
Outcomes for your team
Problems are caught before a customer reports them. Track which issues your team found first. In one team's weekly hardware report, Alloy flagged a joint that was wearing out on one robot, so the team could plan the repair.
Nobody reviews runs by hand. Every run is reviewed; people read only the ones that fail your criteria.
"It took half my job away: all the silly stuff you have to do."
Luke Reid, Robotics Support Specialist. How SwarmFarm is scaling its robot fleet
The team fixes the biggest recurring problem first. The weekly roll-up counts what recurred and on which robots, so priorities follow the data.
Related
- Scenarios & Regression Testing: turn a pattern you found into a check on every new run.
- Validation & Testing: the same recurring report, aimed at a test campaign.
- Root Cause Analysis: what to do with a run that needs a cause.
- Support Agent
Scenarios & Regression Testing
Describe a failure pattern once in plain words. Alloy searches your recent runs for it, checks every new run, and lists each match with the evidence, so you can see whether a fix worked.
Root Cause Analysis
Find out why a robot failed. Ask in plain words: the agent already has your robot data and the tools you connect. It runs its own analysis on the recorded signals, checks every past run, and shows its work.