guide

Guide

Quick reference for what each part of the Demo Combobulator does and when to reach for it. Jump to the section you're working in.

demo environments

Deploy branded Salesforce demos

Demo Environments provision a working Salesforce scratch org preloaded with a branded commerce storefront, example products, and any demo features you enable (skin advisor, cart optimizer, loyalty prompts, etc.). Output: a live URL at <slug>.demo-combobulator.com you can hand to an account team.

  1. 1
    Describe the demo

    Start at New environment. Give it a brand name, pick a vertical, and describe the goal in plain English. Claude maps your description to platforms + features.

  2. 2
    Clone the brand

    Optional: point at a real brand URL to scrape its look-and-feel (colors, copy tone, product examples). Or AI-generate from scratch.

  3. 3
    Review + deploy

    Confirm the config, then Deploy. A scratch org provisions in the background (~5–8 min). The status bar streams the pipeline live.

  4. 4
    Share

    Once status flips to live, the storefront link activates. Share it with teammates via the Share dialog on the demo detail page.

Not seeing features you expect? The intent-interpreter filters features by vertical — e.g. cart optimizer won't appear for a travel demo. If a recommendation looks wrong, check the vertical on step 1.
video recordings

Produce narrated walkthrough videos

Video Recordings turn a series of screenshots into a narrated walkthrough video with your own voice. Build manually step-by-step or let the Smart AI autoscript infer the flow from a list of URLs + a goal.

  1. 1
    Pick a mode

    At New demo video, choose Smart AI (let Claude autoscript from URLs + goal) or Manual Builder (drop screenshots + type scripts yourself).

  2. 2
    Review the script

    Smart AI gives you a reviewable step list — edit any step's narration, reorder, or ask it to revise individual steps before building.

  3. 3
    Build the video

    Build stitches audio + screenshots into an MP4. Progress streams in the activity log. Output is downloadable from the project page.

Sound like you, not a robot. Configure a voice model in Voice Lab first. The builder uses your personal voice for narration; otherwise it falls back to Gemini's generic TTS.
voice lab

Train a voice model that sounds like you

Voice Lab records, trains, and tests a personal ElevenLabs voice model. After training, any video the Combobulator builds uses your voice for narration.

  1. 1
    Record samples

    Open the Recorder and read the prompted paragraphs aloud at your normal demo pace. Aim for 25+ clips (8–10 minutes of audio) — more samples = better fidelity.

  2. 2
    Train

    From the Voice Lab, press Train. Your samples upload to ElevenLabs and a model is fine-tuned. Takes a few minutes; progress streams live.

  3. 3
    Test + set as default

    Type a phrase, hear it back in your voice. Happy with it? Pick it from the dropdown to set it as your default for video narration.

Can I retrain? Yes — record more samples and train again. The old model stays available as a voice option; new ones get new names.