Workflows automate business processes. Policies evaluate rules. AI specialists can help with repeated judgment, outreach, document analysis, and follow-up when the data and allowed actions are clear.
Use workflows for repeatable sequences, policies for rule decisions, and specialists where interpretation or follow-up would otherwise consume operator time.
Use this page when you need to decide:
what to configure, where the data lives, who should see it, and what the next operational action should be.
Give every wait a deadline
A Wait for event step needs a timeout. Choose an amount and a unit, up to 30 days. If no matching event arrives, the workflow takes the timeout path or ends if that path is unconnected.
Example screen with fictional data. Available controls depend on your account and permissions.
This is the vocabulary to use when explaining the concept to a scheduler, payroll lead, client manager, or implementation owner.
1
Trigger
The event that starts the workflow, such as record created, field updated, scheduled time reached, or request submitted.
2
Condition
The rule that decides whether the workflow should continue.
3
Wait
A delay or gated pause, often used before rollups update or before escalation.
4
Action
The outcome: update a record, send a message, assign a task, change access, request approval, or generate a document.
5
Specialist step
An AI-assisted action that reads approved data and performs a scoped task.
1
Workflow patterns
A workflow should be easy to explain in one sentence: when this happens, if these conditions are true, do this next. If that sentence is unclear, the workflow probably needs cleaner data or a narrower scope.
When a user is created, assign onboarding tasks.
When a shift is unassigned, notify eligible workers.
When a DNR field changes, remove restricted locations.
When late count reaches a threshold, move the worker to another tier.
When reimbursement is submitted, route it to admin review.
2
Policies
Policies decide whether an action should be allowed, blocked, flagged, ranked, or optimized. They can evaluate one record or match two records, such as worker-to-shift eligibility.
Blocking policies stop actions such as overtime or missing credential requests.
Flagging policies allow action but warn admins.
Matching policies rank qualified workers for a shift.
Optimization policies prefer lower-risk, lower-cost, or higher-priority matches.
3
AI specialists
AI specialists should be scoped like operational teammates. They need a persona, task instructions, data read permissions, data write permissions, Engage recipient rules, policy access, and file analysis access if needed.
Use specialists for repetitive follow-up, document reading, note parsing, or shift recovery.
Do not use specialists where the business rule is simple enough for a policy.
Give specialists only the data and actions required for the task.
Keep exception paths visible to humans.
Example: Late shift to tier movement
How this shows up in a real Teambridge workflow
Use this as the implementation checklist: each step should produce a visible record, permission, view, or automation.
1
Add a Late field to Shifts.
2
Create a rollup on Users that counts late shifts.
3
Create a workflow triggered when the late count changes.
4
Add a short wait so the rollup can recalculate.
5
If the worker has three late shifts and is Gold, update tier to Silver.
6
Notify the manager or surface the worker in a coaching workspace.