Workload Management
How new interrupts and live sessions are handed to the right member automatically.
Workload Management decides who gets each new piece of work the moment it arrives. When an agent opens an interrupt or starts a Live View session, Vigilator picks a member for it based on the rules you set here - who is online, how much they already have open, and which team the work belongs to - so that nothing sits in the inbox waiting to be noticed.
You'll find it under Settings > Workload Management. Every change saves immediately - there is no save button - and any member can view the settings, but changing them needs the organization.update permission (see access control).
Two surfaces, two configurations
Interrupts and live sessions are very different kinds of work: holding a handful of open interrupts is light, while watching several live transcripts at once is not. Workload Management is therefore configured separately for each surface. The dropdown next to the heading switches between them:
- Inbox interrupts - how new interrupts are assigned.
- Live View - how new sessions are handed to a watcher.
Each surface keeps its own complete set of settings, including its own capacity limit and its own round-robin position. The only setting the two share is what happens when nobody is online, which is configured once under Settings > Notifications (see below).
How assignment works
When a new interrupt or session arrives, Vigilator narrows the organisation's members down to the ones eligible to take it, then picks one:
If nobody is eligible the work is queued: it stays unassigned, visible to everyone with access to the inbox or Live View, and is picked up automatically as soon as somebody becomes eligible again (see The queue). Automatic assignment never overrides a human - operators can re-assign anything at any time, and each automatic assignment is recorded on the interrupt's audit trail.
Settings
Assignment
Auto assign new interrupts / sessions turns the whole engine on or off for the surface. With it off, everything arrives unassigned and is triaged by hand.
The assignment method chooses between the eligible members:
| Method | Behaviour |
|---|---|
| Round Robin (default) | Cycles through members in the order they joined the organisation, so work is spread evenly. Members who are ineligible (offline, at capacity, outside the routed team) are skipped over rather than losing their turn. Each surface keeps its own position in the cycle. |
| Capacity Based | Hands each new item to whoever currently has the fewest open items - unanswered interrupts on the inbox surface, active sessions on the Live View surface. |
Round Robin keeps the count of assignments fair; Capacity Based keeps the amount of open work fair, which matters more when some items take much longer to resolve than others.
Capacity
Maximum interrupts / sessions assignable to a singular member caps how much open work one person can hold. A member at the cap is skipped until something they hold is answered (or their session ends), and if every member is at the cap the work queues.
The cap can be anything from 1 to 1000. The defaults reflect the difference between the two surfaces: 10 open interrupts on the inbox surface, 4 active sessions on Live View. Turn the limit off to let auto-assignment hand a member any amount of work.
Online mode
Only assign to online members stops work being handed to people who aren't there. Every member has a presence status, shown as the dot on their avatar at the bottom of the sidebar, and with online mode on, offline and inactive members are skipped by auto-assignment.
How presence is determined depends on Manually mark presence:
- On (default) - members set their own status with Become away / Become online from the avatar menu in the sidebar (also available from the command menu). A member is online until they say otherwise.
- Off - presence is detected automatically. Vigilator tracks activity while a tab is open and visible - simply watching Live View counts - and a member who has been idle for longer than the Mark inactive after threshold (5 minutes, 15 minutes, 30 minutes or 1 hour) is set to offline. They see "You were set to offline due to inactivity" in the sidebar, and stay offline until they choose Become online - returning activity alone does not bring them back, so that work isn't handed to someone who only glanced at a tab.
In either mode, Become away always wins: a member who has marked themselves away is never treated as online. Presence belongs to the member rather than the organisation, so a member of several organisations has one status across all of them.
Online mode also unlocks the two settings that deal with members leaving: Coverage and When nobody is online.
Coverage
Coverage answers the question "what happens to the work a member was holding when they go offline?". With Reassign work from offline/inactive members on, an offline member's open interrupts (or the active sessions they were watching) are handed back to the queue and redistributed to online members - within the capacity limit, using the surface's normal assignment rules.
The grace period (reassign immediately, or after 10 minutes, 30 minutes or 1 hour away) gives a member time to come back before their work is moved. If they return within the grace period nothing happens. Each surface has its own grace period, so you might move interrupts after 30 minutes but hand off a running session as soon as its watcher steps away.
A few details worth knowing:
- Reassignment leaves an unassigned entry with no actor on the interrupt's audit trail, followed by the new automatic assignment, so it is always clear that coverage moved it rather than a person.
- Work in the manager review queue is never touched - escalated interrupts and sessions stay where they are.
- Coverage requires auto-assignment and online mode to be on for the surface. Without the engine to redistribute it, unassigning a member's work would only strand it.
- Anything the redistribution cannot place (everyone else is at capacity, say) stays queued and is picked up as capacity frees.
Routing
Routing is available on the Inbox interrupts surface only, since it keys off classifications and sessions don't carry one. With Route classifications to specific teams on, you build a list of rules, each mapping a classification to a team with a strictness:
| Strictness | Behaviour |
|---|---|
| Hard limit | Never assign outside the team. If no team member is eligible (all offline or at capacity), the interrupt queues until one is - even if other members are free. |
| Desirable | Prefer the team, but overflow to the rest of the organisation when nobody in it is eligible. |
An interrupt that arrives with a classification is matched against the rules in order and the first match applies. Interrupts with no matching rule - or no classification at all - draw from the whole organisation as usual. You'll need at least one classification and one team before you can add a rule, and a rule whose team has since been deleted is ignored rather than blocking assignment.
Routing only sees the classification an interrupt arrives with. If your agents don't classify interrupts themselves, pair routing with Argus auto-classify so interrupts are categorised - and routed - on arrival.
When nobody is online
The last case is the one coverage can't solve: work arrives and no member at all is online. Because this applies to both surfaces alike, it is configured once, under Settings > Notifications, and only appears when at least one surface has online mode on:
- Leave unassigned (default) - the work queues and waits for the first member to come online.
- Notify the escalation chain - the work still queues, but Vigilator also emails an ordered list of members, one at a time. The first member in the chain receives an email saying how many interrupts are waiting (or sessions are running) with nobody online. If the situation isn't resolved within the Escalate to the next member after timeout (5 minutes, 10 minutes, 30 minutes or 1 hour), the next member is emailed, and so on down the chain.
"Resolved" means somebody came online - signing in is enough, and it drains the queue - or the queued work was cleared. Each step re-checks before sending, so a handled situation stops the chain without a further email. To keep a busy night from turning into a flood, a chain is started at most once an hour per surface.
Note that only the nobody online case notifies. Everyone being at capacity is expected load, not a coverage gap, and simply queues.
The queue
Unassigned work isn't forgotten. Vigilator drains each surface's queue, oldest first, whenever eligibility might have changed:
- When a member comes online - signing in, or choosing Become online - they are immediately handed as much queued work as the rules allow.
- Every minute, as a safety net, to catch capacity freeing up as interrupts are answered and sessions end.
Each queued item goes through the normal assignment rules, so the capacity limit and routing keep holding as the queue empties. An interrupt held back by a hard routing rule doesn't block the ones behind it. Escalated items are left alone - the manager review queue owns them.
Defaults
A new organisation starts with these settings on both surfaces:
| Setting | Inbox interrupts | Live View |
|---|---|---|
| Auto assign | On | On |
| Assignment method | Round Robin | Round Robin |
| Capacity limit | On, 10 | On, 4 |
| Online mode | On | On |
| Manually mark presence | On | On |
| Mark inactive after | 15 minutes | 15 minutes |
| Reassign from offline members | Off | Off |
| Grace period | 10 minutes | 10 minutes |
| Routing | Off | - |
| When nobody is online | Leave unassigned | Leave unassigned |
In other words: work is spread evenly across whoever is online, nobody is handed more than they can hold, and nothing moves or notifies without you switching it on.
Example setups
A small team in one time zone. Keep the defaults. Turn on Reassign work from offline members with a 30-minute grace period so a colleague's interrupts are covered when they leave for the day, and add yourself to the escalation chain so you hear about anything that arrives overnight.
Operators who forget to sign out. Turn Manually mark presence off with a 15-minute inactivity threshold. Nobody gets work assigned to a tab they left open on Friday, and coverage hands it on.
Specialist teams. Create teams such as Security and Finance, create matching classifications, enable Argus auto-classify, and add routing rules - a hard limit for anything that must only be seen by the right team (security actions, compliance reviews), desirable for the rest so it still gets handled when the team is stretched.
High-touch Live View. On the Live View surface, switch to Capacity Based with a capacity of 2 or 3, and set coverage to reassign immediately - a running agent should never go unwatched for longer than it takes the next person to pick it up.