Getting started

From zero to your first screened call, then your first CSV launch.

1. Create your workspace

Sign in or register. The first user becomes the workspace admin. Admins can later invite teammates from Settings > Team and control provider credentials, usage, storage, developer access, and billing-related settings.

You can sign in with your email and password, or use Sign in with Google on the login page. Google sign-in only works for emails that are already registered as a user — if the email isn’t in the workspace, you’ll see “This email is not registered”. It never creates new accounts.

2. Configure providers

If your workspace does run its own keys, make sure these settings are ready:

• Settings > Providers - add or verify the Plivo account used for dialing.
• Settings > AI Services - add credentials for the LLM, STT, and TTS providers your agents use.
• Settings > Storage - configure S3 if you plan to store call recordings.
• Billing > Usage - check monthly minute allowance and wallet balance before large runs.
• Integrations - connect Google Calendar account(s) if you want agents to auto-book interview slots (see the Tools tab in Agents).

3. Add a phone number

Go to My Numbers. Add a Plivo number, keep it active, and mark one number as the default outbound caller ID. This is the number candidates see when campaigns or batches call them. For inbound calling, edit a number and assign it to exactly one agent.

You can also rent a number during a wallet top-up — see Pricing. If your workspace holds several numbers, each agent can be pointed at its own via Outbound number on the agent’s Call tab.

4. Build an agent

Open Agent Setup > New agent. Choose Auto Build to have the welcome, prompt and rating template written from a plain-English description, or Start from Template to begin from a ready-made agent. Either way you land in the builder, where you configure:

• Welcome message — the first line the candidate hears (pre-synthesized for instant playback).
• System prompt — the agent's instructions and the questions it should ask.
• LLM - model, API credential, tokens, temperature, and linked knowledge bases.
• Audio - language, STT provider/model, TTS provider/model, voice, speed, and credentials.
• Engine - Pipeline for modular STT + LLM + TTS, or Gemini Live for speech-to-speech.
• Call - Plivo account, ambient noise, voicemail detection, recording, silence timeout, and call timeout.
• Analytics - rating template, webhook URL, summarization, and extraction prompt.

5. Test it

Start with Web Call in the agent header — you talk to the agent through your browser’s microphone, over the same pipeline a real call uses, and it costs nothing. When that sounds right, use Get a call from agent with your own phone number to hear it over a real line. Check whether the greeting, question order, language, interruptions, and final message feel correct before scaling. See Agents for all three ways to try an agent.

6. Prepare contacts

Use a CSV with at least name and phone (email is optional). Phone numbers must be Indian mobiles in the form +91 followed by 10 digits, with no spaces — for example +916303639014. When you upload, the rows are shown in a preview table and validated: any number that isn't a valid +91 10-digit mobile is flagged, and the Create Draft / Launch buttons stay disabled until every row is valid.

name,phone,email
Asha Rao,+919876543210,asha@example.com
Vikram S,+918765432109,

7. Choose campaign or batch

8. Review results

• Call History - search by call ID, phone, candidate, status, or provider; open a call to read transcript and summary.
• Ratings - review template-based scoring, regenerate a rating, inspect candidate history, and export.
• Recordings - play or download audio if recordings are enabled at platform and agent level.
• Dashboard - monitor call volume, cost, completion rate, funnel progress, and active campaigns.