Employee Wellness
Burnout does not announce itself. It shows up as someone who was excellent becoming merely fine, then quiet, then gone — and the exit interview says “needed a change.” Looking back, the signs were all there and all boring: they had not taken a full week off in fourteen months, their team had been down two people since spring, and they had been on call every third week since the reorg. Nobody saw the pattern because no one person could see all of it at once, and asking each individual how they are doing reliably produces “fine.”
The signal is structural and it is visible at the team level. Skynet tracks the indicators you decide are meaningful and tells a human when a team’s picture has changed. Where that line sits matters enormously, so it is a line you draw, not one the agent assumes.
How it works
Decide what gets watched
This is the first step because it is the important one. Pick the indicators you are comfortable tracking — unused leave balances, sustained vacancy on a team, on-call rotation load, pulse survey trends. Anything not on that list is not looked at. The agent monitors the workload structure, not people’s behavior.
Report at the team level
Findings are about teams and functions, never individuals. The agent does not produce a list of at-risk employees, because that is a different product with a different set of consequences. It tells you that one team’s leave usage has collapsed and its vacancy load has doubled — which is enough to act on.
Look for the sustained pattern
One heavy month is a heavy month. The agent watches for the conditions that persist: a quarter of carrying two open roles, a rotation that has not rebalanced since spring, a downward pulse trend that has not recovered. Duration is what separates a crunch from a problem.
Hand it to a person
Every finding goes to the people team as something for a human to look into, with the specific pattern shown. The agent does not message an employee, does not tell a manager someone is struggling, and does not draw a conclusion about anyone’s state. The follow-up is a conversation, and conversations are human work.
Build it from a prompt
You set the boundaries explicitly, and the agent stays inside them.
You find out that a team is running on empty while it is still fixable with a hire or a rebalanced rotation — not after two of them quit. What the agent gives you is a place to look and a conversation to have. Trust is the entire asset here, which is why the scope is narrow by design, the reporting stops at the team, and a person always makes the first move.