Docs · Agents

The demo agent.

The demo agent builds the demo. Each prospect gets a live AI assistant preloaded with their business, picked from one of four active surfaces (receptionist, lead reactivation, website chat, missed-call text-back), reachable from a private link the LinkedIn or email agent drops wherever you reference {{demoLink}} in the sequence, no fixed message position. The prospect clicks, talks to their own AI in the browser for a few minutes, and shows up to the call already half-sold.

Quick facts

Channel
Web-RTC voice + on-site chat (browser, no phone)
Auth
Platform Vapi key (no per-operator setup)
Provisioning
On-demand at message-render time
Demo TTL
14 days from the send that carries the link, then the expire-demos cron tears the assistant down
Templates
Receptionist, lead reactivation, website chat, missed-call text-back

How a demo gets to the prospect

  1. You build a demo on the demo agent's profile at /dashboard/team/theo, picking either or both voice agent types (Receptionist, Lead Reactivation) plus an optional on-site chat widget. One branded demo link per prospect. Or just ask Ciela in the Team tab to build a one-off demo for a single prospect (give it the website and demo type in one line) and it builds it in chat, no form needed.
  2. When the LinkedIn or email agent is about to send a message that references {{demoLink}}, the demo agent provisions a per-contact Vapi assistant. The agent's prompt is preloaded with the prospect's company, owner name, and any niche notes you've saved on the campaign. If the contact has a website URL on file, the demo agent also pulls in business context from the prospect's site so the agent sounds like it actually works there instead of guessing.
  3. The prospect clicks the link, lands on a public hub at /demo/{slug} showing every demo in the kit, picks one, and is in a browser-to-AI conversation in under three seconds.
  4. The demo agent never books, transfers, or writes to your calendar from the demo, the prospect comes back to the email or DM thread to book with you directly.
  5. 14 days after the demo is provisioned, the expire-demos cron deletes the assistant on Vapi and marks the demo expired in the database. The link still resolves but returns an already-expired view.

What it will not do

  • Won't re-demo someone already engaged. If the LinkedIn or email agent already got a reply on the contact inside the same flow, the demo agent skips the provision step, that's the omnichannel agent's coordination at work. Same cross-surface conversation-stop gate the other agents share.
  • Won't book or take actions on the demo. The Vapi assistant is provisioned with no tools and no server URL, the prospect's AI can talk, but it can't write to your calendar, your CRM, or anything else.
  • Won't pretend the demo is a real deployment. The demo agent identifies itself as a sample and tells the prospect this is a preview of what you'd build for their business. Honest framing, no bait-and-switch.
  • Won't keep the assistant alive past the TTL. 14 days after provision, the demo agent cleans up. If a prospect clicks an expired link they get a graceful expired view, not a broken connection.

Where it fits

The demo agent lives as a separate first-class agent, not as a flow step type. The LinkedIn and email agents reference demos through the {{demoLink}} template variable that resolves to a per-contact URL at message-render time. That means one campaign can drop the demo in an email AND in a LinkedIn message without provisioning twice, and the prospect always lands on the same demo regardless of which channel they clicked from.

What it costs

Demos run on your personalized-demo allowance, a separate monthly bucket from your AI credits. A personalized demo is one unique prospect hub (one prospect or company) per month, and every plan starts with 500 of them. Demo generation, prospect demo chat, and prospect demo voice are all included under that allowance, never billed to your AI credits.

  • One demo, one count. Provisioning a prospect hub counts as a single personalized demo for the month, no matter how many times the prospect later chats or talks to it.
  • Demo chat and voice are included. Prospect turns on the floating chat widget and live voice calls run under fair use of your demo allowance, not your AI credits. Your operator-directed AI (Ciela chat, AI writing, lead sourcing, outreach generation) is the only thing that draws credits.
  • Need more demos? Stack a monthly demo-volume add-on (Demo Starter, Demo Growth, Demo Scale) to raise the allowance. It adds demo capacity only, never extra AI credits.

Customizing the demo page

The agency name, logo, brand color, and per-prospect CTA shown on every demo page (single demos and the prospect hub) come from your Brand settings. Edit it once, every demo the demo agent ships uses it. The CTA supports token substitution ({{first_name}}, {{business_name}}, {{owner_name}}) so the conversion ask reads personalized to whoever the prospect is.

How to put it to work

Three ways to put the demo agent to work, depending on whether you want a standalone demo, one wired into an outbound flow, or the whole niche staged from a single sentence in Ciela chat.

Path A: standalone kit (direct share)

  1. Open /dashboard/team/theo and click New demo.
  2. Pick either or both voice agent types (Receptionist, Lead Reactivation) and optionally add the on-site chat widget.
  3. Fill in the prospect's business name, owner, primary service, and website (required for chat). Hit Build.
  4. The demo agent provisions every demo and gives you one branded link at /demo/{slug}. Copy it, drop it in a DM, an email, wherever you talk to that prospect.
  5. Track views and last-accessed timestamps per demo on the same demo agent profile page. High views + low completion means the landing page is doing the work but the demo itself has friction.

Path B: flow-embedded (automated across the agents)

  1. Reference {{demoLink}} in any message body in a LinkedIn, email, or omnichannel campaign. Put it wherever fits the cadence, the demo agent doesn't enforce a position.
  2. Launch. The demo agent provisions per contact as the LinkedIn and email agents send. You don't manage assistant lifecycles, the expire-demos cron does that once each demo's 14-day TTL elapses.
  3. Watch provisioned, clicked, and engaged counts on /dashboard/team/theo.

Path C: the Acquisition Run (one sentence in Ciela chat, Ciela AI)

  1. In the Team tab, tell Ciela one line like “sell AI agents to med spas, find the owners, write the outreach, and build each one a demo they can try.” This runs on demos, so it is included with Ciela AI.
  2. Ciela stages the whole machine at once: a pending owner lead list, an outreach campaign parked until the owners land that already carries the selected demo and {{demoLink}}, a demo batch, and one live sample demo you can click right now. Ciela picks the channel for the accounts you have connected: an omnichannel flow when both email and LinkedIn are connected, standalone LinkedIn when only LinkedIn is, or email when that is your connected channel. Whichever it is auto-starts when the owners land.
  3. Only the sample is live immediately. Sourcing the owners takes about 24 to 48 hours, and each prospect's real demo provisions at send time when the outreach goes out, not up front.

Keep reading

Need help?

Ciela can walk you through any step. Or write to support@ciela.ai and we'll jump on the connection with you.

Ask Ciela