Agents
In AssureAgent vocabulary, an agent is the caller persona — the voice / text on our side of the conversation. The thing on the other side (your product) we call the target agent, to keep ambiguity out of the docs.
What an agent owns
An agent in your workspace owns the delivery characteristics of the simulated caller:
| Setting | What it controls |
|---|---|
| Voice | The voice timbre. Picked from a curated catalog of available voices. Voice ships with the agent — same voice on every call this agent makes. |
| Responsiveness | How fast the caller starts speaking after the target stops. Lower = thinking pauses; higher = quick replies. |
| Interruption sensitivity | How easily the caller cuts in if the target trails off mid-sentence. |
| Max call duration | Hard hang-up cap in milliseconds. We end the call automatically at this limit no matter where the conversation is. |
| Ambient sound | Optional background ambience (office hum, coffee shop, none) layered into the audio. Useful for stress-testing the target’s noise robustness. |
What an agent does not own: the persona content (objectives, opening line, behaviors, edge cases). That lives on the scenario so a single agent can play many different personas across many scenarios.
How many agents to create
Most workspaces start with one default agent and grow only as they need different delivery profiles:
- One agent that’s calm + slower for empathy-sensitive scenarios.
- One agent that’s quick + interruptive for impatient-customer flows.
- A separate voice for a Spanish-language test set.
If two scenarios just need different content but the same delivery, share a single agent.
Editing an agent
Edit any field anytime — the change applies to future test calls. Calls already running keep their starting settings. If the agent is attached to scenarios that use custom functions, the edit also propagates to the per-scenario specialized clones we maintain in the background, so max_call_duration (and friends) stay in lock-step.
Where to find them
Left nav → Agents. Each row in the list shows name, voice, max call duration, and how many scenarios the agent is currently attached to.