Present the Allianz journey in a Scene¶
Last reviewed: 2026-10-14
In Agent Studio, open Agents → Allianz · Producer Services journey (4911). It opens in Director. This saved Scene combines the self-service phone journey and the Allianz Support Services Salesforce experience.
Select Play recording for the complete 6 minute 22 second walkthrough, or select a chapter and use the playback slider to rehearse a section. Preview enlarges the stage. Export creates an MP4 using the same saved voices and screen timing. A completed 1080p recording is already available in Exports.
The recording follows four chapters:
- Self-service · Verify and compare withdrawals — David asks for values, completes Face ID, opens secure chat, reviews Helen's contract and withdrawal options, and requests a representative.
- Support Services · Confirm and share — an incoming ring accompanies the handoff. Sanjay accepts and reviews Confirm with David. After David's death-benefit question and the producer guidance, he says: “Yes, I can help explain that. I'll also send this information to your phone so you can follow along on screen.” Contract Snapshot stays in the center.
- David’s phone · Death Benefit Impact — after Send to Web Chat, the button disappears from below the benefit card; the bordered card stays unchanged. The four-second cutaway centers David's phone without AI Activity. His existing chat appears briefly, then the same estimate and qualification slide into view.
- Support Services · Explain, illustrate and wrap up — back in Salesforce, Sanjay begins “Helen’s contract includes the…” and explains the benefit. He then opens the form, confirms the inputs, submits, shares the signature follow-up task and reviews the normal Call Wrap Up modal.
The assistant uses Clara, David uses Brian, and Sanjay uses Jeremy in the October 14 refresh. The clarity interruption and English Enhancement UI are absent. Sanjay's voice stays consistent throughout the call. Explain the live normalization technology after the walkthrough.
Use Script to edit lines and Voices to change the cast. Regenerate audio for changed speech before exporting. Screen cues follow their associated lines. The Salesforce chapter's Stage controls hold its branding, record details, workspace sections and Agent Assist component.
Suggestions appear about 1.6 seconds after David's answer. Sanjay has at least one second to read them before responding. Self-service replies also include short latency pauses. A separate cursor layer moves to controls and shows clicks. At the end of the call, Sanjay responds after a 1.2-second pause while the follow-up sharing and toast play alongside his farewell. The normal wrap-up modal follows. Phone clocks, the call timer and Messages use consistent iPhone-style typography in playback and export, and the caller name appears in full. The Salesforce view follows the deployed Support Services components: Arial for the Case, workspace tabs, Contract Snapshot, illustration form and Assist heading; the native system font for navigation and the producer profile. The Snapshot has its bordered layout and larger detail text. Producer Summary uses the compact sparkle heading and list spacing from Salesforce.
The Case title includes Producer Request, with Edit and Mark Complete buttons above the status Path. Completed steps show checkmarks. The workspace includes Commissions, and Current Case shows the call summary. The Assist panel has separate Agent Assist and Transcript tabs.
The illustration form fills progressively, with separate amount and frequency controls. Allocation and purchase payments appear after David agrees. The confirmation checkbox stays clear until the modeling readback and his final agreement; Submit Request then becomes available. The receipt represents a saved request with document delivery pending. Submission restores Contract Snapshot with an Illustration Request Submitted notification above it. Share follow-up task displays a success toast. Wrap-up uses Policy Inquiry and finishes with a closed Case.
This is an authored Scene. Playback does not call Salesforce, place a real phone call, generate an insurer document, or run live accent conversion. For the connected demo, use Allianz DCC Support Services and the live self-service agent.
Ringtone and voice comparisons¶
Open Compare.html in the extracted comparison ZIP. The page is titled Allianz AI Producer Support Experience and offers, in order, Full Voice Replacement, Basic Accent Reduction, and Original. All three videos include the corrected phone typography and Salesforce component layout, fonts and spacing. They share the updated Scene's 6:22 timing, so switching versions keeps your playback position. The highlight shortcuts jump to a talking point and pause with its presenter notes. The page has no explanatory subtitle, version description, footer, or download link.
The official Scene and comparison package both include the shorter farewell pause and the updated saved voices. Use the latest revision 110 export.
The incoming Salesforce call uses the recorded Standard phone ring. New Scenes can reuse it through Insert moment → Sound effect; see Add the standard phone ring.
The comparison videos use the same screens and caller voice. Full Voice Replacement uses the produced Scene dialogue and voices. Basic Accent Reduction uses the recorded human agent with Krisp noise cleanup and accent normalization. Original uses the unprocessed recorded human agent. The recorded versions preserve the original wording. The separate Krisp + ElevenLabs rendition is excluded from the presentation page and ZIP.
Below the player, Estimated cost & added delay estimates a planned live deployment using ElevenLabs v4 Turbo + OpenAI or Krisp. Adjust Call length and Agent speaking; the default ten-minute call with 50% agent speech costs about $0.188 for full replacement and $0.100 for Krisp. The 50% share is a planning assumption, not an industry average. Include separate transcription raises the replacement estimate to about $0.218; otherwise it assumes an existing usable transcript.
Open Assumptions, calculation & latency basis for unit prices, token budgets, sources and exclusions. The replacement delay is a provisional 3.5–4.5 seconds with a three-second phrase buffer, including 0.5–1.5 seconds after capture. Krisp's approximate 0.3 seconds comes from offline clip alignment. Neither figure is a measured live-call guarantee. The existing comparison audio uses the saved Scene voices; the estimate does not regenerate it with Turbo.
Present with highlights¶
Select Play demo, or use the video's native play control. Pause at highlights is on by default. The player pauses at each point below, shows your talking points in the side panel, and waits for Continue. The main play control and the native video control also resume playback.
Before the first pause, the panel shows Independent Producer Journey: “David Chen is calling on behalf of his client Helen to explore her annuity surrender and withdrawal options.” After each Continue, the panel clears its heading and bullets, leaving the Next… message and playback controls at the bottom. The next set of talking points appears when playback pauses again. Each highlight has two bullets. Authentication covers on-device biometrics and integration with existing digital experiences. Multimodality covers expanding what AI can handle on the phone and helping visualize financial scenarios.
| Pause | Talking point |
|---|---|
| 0:51 | Authentication |
| 1:40 | Multimodality |
| 2:33 | Context Handoff & Producer Summary |
| 3:17 | AI Coaching & Multimodality |
| 4:55 | Automatic Form Filling |
| 5:55 | Proactive Suggestions |
| 6:07 | Call Wrap-Up |
The October 14 voice refresh moves the original stops to their corresponding moments in the new delivery. Multimodality adds a separate pause at exactly 1:40.
Restart demo returns to the beginning with highlights enabled. Clear Pause at highlights for continuous playback. Seeking forward skips earlier stops; seeking backward makes later stops available again. Use Fullscreen in the side panel to keep the video, talking points, and Continue button together.
Present the accent comparison¶
After the main recording, select Play accent demo. It is also available at the top of the page at any time. This sequence uses the same video timeline:
| Interval | Audio |
|---|---|
| 3:27–3:42 | Original recorded voice |
| 3:42–3:57 | Basic Accent Reduction — Krisp |
| 3:57–4:12 | Full Voice Replacement — Jeremy |
The player changes audio without reloading at either transition and pauses at 4:12. Each voice still plays for 15 seconds. The side panel shows the active voice and its estimated ten-minute call cost: $0.10 for Krisp or about $0.19 for full replacement at 50% agent speech with an existing transcript. The full replacement estimate is for the planned v4 Turbo + OpenAI workflow; the soundtrack remains the existing produced voice. The speaking-share and transcription controls also update these side-panel costs. Its call length stays fixed at ten minutes. The price card shows the amount and Estimated per 10-minute call only; rounding details, assumptions beneath the caption, and the end-of-demo voice-sequence text are removed. Full assumptions remain available in the calculator below the player.
You can pause and continue the accent sequence at any time. Replay accent demo
starts it again; Continue recording resumes after 4:12 with the produced Scene
voice. Selecting a voice manually exits the automatic accent sequence and keeps
the current position. The ZIP includes the three comparison videos plus
allianz-accent-demo.mp4; keep all four MP4s beside Compare.html.