# Prompt Management

**Navigation:** Admin → Architect → Prompts
**Context:** Part of Architect administration — managed alongside flows, not under Routing

---

## What Are Prompts?

Prompts are **audio or text-to-speech messages played to customers during interactions** inside Architect flows. Every piece of audio a caller hears — welcome greetings, menu options, hold messages, closure announcements — is a prompt.

---

## Two Types of Prompts

| Type | Description | Editable? |
|---|---|---|
| **System Prompts** | Pre-built by Genesys for generic use (dates, numbers, days, months, standard phrases) | Cannot be renamed or deleted |
| **User Prompts** | Custom prompts created by administrators for org-specific messaging | Fully configurable |

> ℹ️ System prompts are used internally by Architect for things like reading back times, dates, and digits. You do not create or manage them — you only create User Prompts.

---

## Prompt Content Options

Each User Prompt can use one or both content types:

| Option | Description |
|---|---|
| **Text-to-Speech (TTS)** | Type the message text — Genesys synthesizes it using the org's configured TTS engine |
| **Uploaded Audio (WAV)** | Upload a professionally recorded audio file |

> ✅ **Best practice:** Use TTS for dynamic or frequently changing messages. Use uploaded WAV for polished, permanent messages like main greetings where audio quality matters most.

---

## Prompt Naming Rules

| Rule | Detail |
|---|---|
| No spaces | Use underscores instead (e.g., `US_Support_WelcomePrompt`) |
| No special characters | Letters, numbers, and underscores only |
| Cannot start with a number | Must start with a letter |
| Must be unique | No two prompts can share the same name in the org |

---

## Creating a Prompt

1. Admin → Architect → **Prompts**
2. Click **Add**
3. Enter the **Prompt Name** (follow naming rules above)
4. Optionally add a **Description**
5. Click **Create Prompt**
6. In the prompt editor:
   - For TTS: type your message text in the **Text-to-Speech** field
   - For audio: click **Add Audio** and upload a WAV file
7. Click **Save**

[![](https://wiki.tinod.net/uploads/images/gallery/2026-03/scaled-1680-/8WCv3jaju3bZewPS-image-1773359060589.png)](https://wiki.tinod.net/uploads/images/gallery/2026-03/8WCv3jaju3bZewPS-image-1773359060589.png)

[![](https://wiki.tinod.net/uploads/images/gallery/2026-03/scaled-1680-/ZntnkM3ayxjKlNCH-image-1773359066068.png)](https://wiki.tinod.net/uploads/images/gallery/2026-03/ZntnkM3ayxjKlNCH-image-1773359066068.png)

---

## Managing Prompts

| Task | How |
|---|---|
| Edit a prompt | Admin → Architect → Prompts → click prompt name |
| Update TTS text | Open prompt → edit text field → Save |
| Replace audio file | Open prompt → Add Audio → upload new WAV → Save |
| Add a language variant | Open prompt → add TTS or audio for additional supported language |

---

## Multi-Language Prompts

A single prompt can have content defined for multiple languages. Architect selects the appropriate language variant at runtime based on the flow's language configuration. If a variant for the active language doesn't exist, Architect falls back to the default.

---

## Using Prompts in Architect Flows

Prompts are referenced inside flows using the **Play Audio** action (or within Menu actions for IVR options). The flow does not embed the audio — it references the prompt by name from the central library.

**Result:** Updating a prompt in the library automatically updates it everywhere it is used, without republishing flows.

```
Architect Flow
      ↓
Play Audio action
      ↓
References prompt: US_Support_WelcomePrompt
      ↓
Prompt library serves TTS or WAV
      ↓
Customer hears message
      ↓
Flow continues
```

---

## Common Prompt Use Cases

| Prompt Type | Example |
|---|---|
| Welcome Greeting | "Thank you for calling Customer Support." |
| IVR Menu | "Press 1 for Sales, 2 for Support, 3 for Billing." |
| Queue Hold Message | "All agents are currently busy. Your call is important to us." |
| Estimated Wait | "Your estimated wait time is approximately [X] minutes." |
| Holiday Closure | "Our offices are closed for [holiday]. We will reopen on [date]." |
| After Hours | "You have reached us outside our business hours." |
| Maintenance | "We are currently performing scheduled maintenance." |

---

## Naming Convention

| Format | Example |
|---|---|
| `<Region>_<Dept>_<Purpose>` | `US_Support_WelcomePrompt` |
| `<Dept>_<Function>_Prompt` | `Support_MenuPrompt` |
| `<Region>_<Service>_Announcement` | `EU_Service_HolidayAnnouncement` |

---

## Troubleshooting

| Issue | Cause | Fix |
|---|---|---|
| Prompt not playing in flow | Prompt not referenced in the Play Audio action | Open flow → verify the prompt action points to correct prompt |
| Audio playback failure | Incorrect file format | Re-upload as a valid WAV file |
| TTS not working | Text formatting issue (special characters, symbols) | Review and clean the TTS text content |
| Prompt change not reflected in live calls | Flow not republished after prompt update | Prompts update without republish — if issue persists, check the flow's prompt reference is correct |
| Wrong prompt playing | Incorrect prompt name referenced in flow | Update the Play Audio action to reference the correct prompt |

> ℹ️ Unlike flow changes, **prompt content updates (TTS text or audio replacement) take effect without republishing the flow**. However, if you rename a prompt or create a new prompt to replace an old one, you must update the flow reference and republish.

---

## Troubleshooting Checklist

| Check | ✓ |
|---|---|
| Prompt created in Architect | ☐ |
| TTS text entered or WAV file uploaded | ☐ |
| Prompt saved | ☐ |
| Play Audio action in flow references correct prompt | ☐ |
| Flow published (if flow changes were made) | ☐ |
| Test call/interaction confirms correct audio | ☐ |

---

## Key Facts for Exam / Interview

| Question | Answer |
|---|---|
| Where are prompts managed? | Admin → Architect → Prompts |
| What two content types can a prompt have? | Text-to-Speech (TTS) and uploaded WAV audio |
| Can system prompts be deleted or renamed? | No |
| What file format is required for uploaded audio? | WAV |
| Do prompt updates require republishing the flow? | No — content updates are live immediately; only flow reference changes require republish |
| Can one prompt serve multiple languages? | Yes — add language variants within the same prompt |

---

## See Also

- **Architect Overview** — where prompts are used inside flows
- **Call Flow Components & Basics** — the Play Audio and Menu actions that reference prompts
- **Organization Settings → Global Settings** — Default TTS Engine configuration