Scenarios

A scenario is one situation you want to test against your target agent. Examples:

• “Frustrated cardholder disputing a duplicate charge.”
• “First-time patient scheduling an appointment.”
• “Caller pressing 1 for English then 2 to file a claim.”

A scenario in your workspace owns the content of the test (what the caller is trying to do, what they know, how they react) plus runtime configuration (which target number to dial, which agent to use, success criteria).

Two design styles

You pick the design style when you create the scenario; it’s the thing that most shapes how you’ll author.

Graph

A visual canvas of nodes and edges. Each node is a turn (caller speaks, target speaks) or a control flow element (branch, service call, end). You drag, drop, and connect.

Best when:

• The conversation is deterministic — you know exactly what should happen at each turn.
• You’re testing IVR menu navigation with specific digit-press sequences.
• You’re integrating with downstream service calls that have to fire at a specific point.

Free-flow

A natural-language description of the situation. The platform converts the description into a structured caller persona at runtime — opening line, objectives, behaviors, escalation triggers, edge cases.

Best when:

• The conversation is open-ended — many valid replies at each turn.
• You want to cover a space of related conversations from one piece of authoring.
• You want the caller to invoke custom functions — those are exclusive to free-flow.

You can switch design style after creation, but the existing content (graph nodes vs description text) doesn’t auto-translate.

Scenario fields

Every scenario, regardless of style, owns:

Field	Required	Notes
Name	yes	Short label; shown in lists and reports.
Type	yes	outbound or inbound. Affects which phone fields are required.
Agent	yes	The caller persona agent used during runs.
Target phone number	for voice	The number we dial (outbound) or accept calls from (inbound).
Outbound phone number	for outbound voice	Which provisioned number we place calls from.
Description	for free-flow	The natural-language situation description.
Custom functions	optional, free-flow only	Caller-side tools. Full guide.
Success criteria	optional	Free-text rubric used to evaluate runs. Auto-generated if you don’t write one.
Labels	optional	Tags for filtering and aggregation in reports.

Scenario lifecycle

1. Create — name + style + agent + target.
2. Author — fill the graph or write the description.
3. Generate test paths — expand into runnable narratives. Learn more →
4. Run — kick off test runs.
5. Review — read the reports, iterate on the scenario or the target.
6. Schedule (optional) — recur the runs nightly or post-deploy.

Where to find them

Left nav → Scenarios. Filter by label, type, or design style. Click any scenario to open the designer or jump straight to its paths and reports.