Browse Documentation

Introduction

RevRag builds AI voice agents for banking, lending, insurance, and wealth apps. They guide your users inside your app and call them on the phone.

RevRag's agents work like a personal relationship manager for each of your users:

  • In your app, the agent talks with users by voice and knows which screen they're on. On supported SDKs, it can also act on the screen (tapping, typing, and scrolling) to walk a user through onboarding, KYC, or an application.
  • On the phone, the agent calls the users you choose, for example to recover an abandoned loan application, renew an insurance policy, or activate a new credit card.

These docs are for the engineers who connect RevRag to your product. They cover the in-app SDKs, the Campaigns API for outbound calls, the webhooks that send each call's result to your backend, and RevRag's security and data-handling practices.

Choose an integration

In-app agentCalling agent
Reaches usersInside your app, while they use itOn their phone, through outbound calls
Good forOnboarding and KYC, in-app sales, feature adoption, user supportAbandoned-application recovery, policy renewals, card activation, investment reactivation
You buildAn SDK integration in your appCalls to the Campaigns API from your backend
ResultsSDK events while the call is live, and a call.ended webhook after it endsRecord status from the API, and a call.ended webhook after each call

You can use both, and they can share what each user has already told them. See How it works.

How it works

You identify the user

Your app initializes the SDK with your API key. After the user logs in, it sends their ID as app_user_id. For phone calls, your backend instead pushes a record to a campaign: a phone number, any variables the agent uses, and the same app_user_id.

The agent talks with the user

In your app, the user starts a live voice call from the agent's floating button, and the agent follows along as they move between screens. By phone, the campaign dials each record within its calling window, retry rules, and concurrency limit.

Your backend gets the result

Once webhooks are turned on, RevRag sends a signed call.ended webhook to your server when each call ends. It carries the call's status, summary, transcript, recording URL, extracted variables, and your app_user_id.

Context carries across channels

RevRag keys conversation memory on app_user_id. When your app and your campaign records send the same ID, each agent picks up the context of the user's conversations with the other.

Add the in-app agent

Every SDK guide follows the same core steps: install the package, initialize it with your API key from the RevRag dashboard, add the agent to your app, and identify the user after login.

PlatformPackage
React Native@revrag-ai/embed-react-native
Flutterembed_flutter
Android (Kotlin)ai.revrag:embed-android
iOS (Swift)RevragEmbed
React, including Next.js and Preact@revrag-ai/embed-react
Angular@revrag-ai/embed-angular

If your mobile app opens your website in a WebView, follow the React Native WebView or Flutter WebView guide to give the agent microphone access.

Start a calling campaign

  1. In the RevRag dashboard, create and start a campaign. The campaign sets the agent, calling schedule, retry rules, and concurrency limit. Copy its campaign_id from the campaign's Settings tab.

  2. Push records from your backend, with your workspace API key in the X-API-Key header. Each accepted record returns a run_id.

    curl -X POST "https://api.revrag.ai/v1/campaigns/$CAMPAIGN_ID/records/push-bulk" \
      -H "X-API-Key: $REVRAG_API_KEY" \
      -H "Content-Type: application/json" \
      -d '{
        "records": [
          {
            "phone_number": "+919876543210",
            "variables": { "first_name": "Riya" },
            "app_user_id": "user_12345"
          }
        ]
      }'
  3. Track each record by its run_id with Get Record Status, or receive a call.ended webhook when each call ends.

The Campaigns API reference documents every endpoint, field, and error.

Webhooks and security

Build with AI tools

  • Integration prompts: the React Native, Flutter, Android, React, and Angular guides each have a prompt for AI coding agents such as Copilot, Cursor, and Claude Code. Open it with the llms.txt button at the top of the guide, then paste it into your coding agent in your app's repository.
  • Markdown pages: add .md to any page's URL, or select View as Markdown.
  • Whole docs: /llms.txt indexes every page, and /llms-full.txt has all of them in one file.
  • Ask AI: select Ask AI at the top of the page to ask a question about these docs.

Get help