Orchestration Agents
An Orchestration Agent coordinates your Call and Chat Agents to reach a specific goal, and stops when the goal is reached or a guardrail is hit.
Rebook missed appointments
An Orchestration Agent is a job you can repeat. Each run works toward one goal. Goals can target one person or many.
You start a run
Your app kicked off a run with any required inputs like the contact's information and their time zone.
It executes the best step based on your configuration
Based on your goal, it triggers calls, texts, or tools using your approved agents.
It finds the best time
It can wait until a better time, such as the next morning, and then try again.
It tracks and hits milestones
Every agent conversation ends with a structured result and updates the milestones your agent has reached. The run uses that context to decide the next best step.
It keeps going, inside your guardrails
It keeps working until the goal is reached or it hits a limit you set.
It finishes with an outcome
The run ends as succeeded, failed, or cancelled, with a result shared with you in your schema.
Rebook missed appointments
Orchestration Agent
Your Orchestration Agent stays within your guardrails, every time.
Check the milestone a run has reached with GET. When it ends, it reports its status and what it cost. Cancel a run at any time.
Rebook missed appointments
ExampleStatus
Milestones
Cost of this run
Some jobs need a text, a wait, and a call. Some need calls to several businesses. Here are a few examples.
When a patient misses a visit, the run texts them to offer a new time. If the chat ends with no answer, it waits a few hours and calls once. It stops when the patient gives a preferred time. Then, the front desk confirms the booking.
The run calls to confirm. If no one answers, it tries again later, inside the calling window and the call limit.
The run calls several businesses in turn and asks whether they have an item. It stops at the first one that does, or at the call limit.
Ask the candidate for available times by text. Call the interviewer to check those times. Report a shared time to your team.
Call local providers to check availability for a repair. Text the customer the options. Report their preferred provider and time to your team.
Text a new lead to ask what they need. Call if they do not reply. Share the result with the sales team.
Set up an Orchestration Agent yourself in the portal or through the API, or describe the job to Butler and it drafts one for you. Butler can then review runs and adjust the agent.
When a patient misses an appointment, text them to rebook. If they don't answer, call once in the afternoon. Never more than two calls.
The API still uses the earlier name: /v1/workflow-agents, /v1/workflow-runs and workflow_run.ended.
curl https://resia.work/v1/workflow-runs \
-H "Authorization: Bearer $RESIA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"workflow_agent_id": "YOUR_WORKFLOW_AGENT_ID",
"inputs": {
"patient_first_name": "James",
"phone_number": "+16175550188",
"missed_appointment_time": "Today, 10:00 AM",
"time_zone": "America/New_York"
}
}'Describe the goal to Butler, set your guardrails, do a test run, and push live.