One of the questions we hear most often from new customers: "How long does it actually take to go live?" The honest answer is 7 days for a standard deployment — and under 3 days for teams with clean data and a clear use case. Here is the exact process.
Before You Start: What You Need Ready
- A contact list (CSV with phone numbers, names, and any relevant data fields like outstanding amount, policy number, etc.)
- A clear call objective (collections, lead qualification, appointment booking, support — pick one for the first deployment)
- Designated point of contact for the first week (someone who can review call recordings and give feedback)
- Your telephony setup: either a Twilio account, Airtel Business account, or any SIP-compatible provider
Day 1–2: Account Setup and Knowledge Base
Create your Agni account
Sign up at ravan.ai and select your plan. For most first deployments, the Growth plan (₹9,999/month) covers everything: 10 AI agents, 25 concurrent calls, all Indian languages.
Build your knowledge base
The knowledge base is what Agni draws on to answer questions. Upload:
- Your product/service documentation (FAQs, pricing, policies)
- Compliance constraints (call window rules, consent language, escalation triggers)
- Industry-specific data (loan terms, policy details, property specs — whatever is relevant)
Pro tip: Start with a focused knowledge base. 20 well-written FAQs outperform 200 loosely formatted documents. You can expand it after the first week of live calls reveals the actual questions callers ask.
Day 3: Call Flow Design
Agni uses a conversation flow — a structured guide that defines the call's opening, key questions, branching logic, and closing. For standard use cases (collections, lead gen, appointment booking), we provide templates.
Key decisions in call flow design:
- Opening: How does Agni introduce itself? Does it identify as an AI upfront or use a first name?
- Primary objective: What is the single most important outcome of this call?
- Objection handling: What are the 5 most common objections in your industry, and how should Agni respond?
- Escalation trigger: When does Agni hand off to a human agent?
- Closing: What action does the caller take at the end — payment link, appointment booking, callback request?
Day 4: Language and Voice Configuration
Select your primary language and fallback languages. For most North Indian businesses, the stack is: Hindi primary → Hinglish fallback → English if needed. For South India, configure Tamil/Telugu/Kannada as primary.
Agni's language detection is automatic — if a caller responds in a different language than the opening, Agni switches. But you need to configure which languages are permitted for your deployment.
Day 5: Telephony Integration
Connect your telephony provider. If you are using Twilio or Airtel, the integration is a 15-minute configuration in the Agni dashboard — API keys, phone number selection, and webhook endpoints. For custom SIP setups, our team handles the configuration.
Test calls before go-live:
- Run 10–20 test calls with internal team members playing the role of real callers
- Test edge cases: caller speaks in unexpected language, caller asks off-script question, caller is hostile
- Verify recording and logging is working correctly
- Confirm compliance rules are enforcing correctly (call window, consent capture)
Day 6: Pilot Launch (Small Batch)
Do not launch to your full contact list on day one. Select 100–200 contacts for a pilot batch — representative of your full list but small enough to review manually.
After the pilot runs, review:
- Connection rate (what percentage of numbers connected)
- Completion rate (what percentage of connected calls reached the objective)
- Drop-off points (where in the flow do most calls end without completion)
- Transcription accuracy (spot-check 20–30 call transcripts for mis-transcriptions)
Day 7: Tune and Scale
Based on pilot data, make targeted adjustments to the call flow and knowledge base. Common Day 7 changes: adjust opening tone, add objection handling for an unexpected objection, fix a specific Hinglish phrase that was being mis-transcribed.
After adjustments, scale to your full contact list. Most customers are at 500–1,000 calls per day by end of Week 1.
"We went from signing the contract to 800 live calls per day in 6 days. The Agni team was available throughout — every question was answered same day." — Operations Head, Mumbai EdTech
Want help with your deployment? Every Growth and Scale plan includes dedicated onboarding support — a Ravan.ai team member works with you through the entire 7-day process. Book your demo at ravan.ai to start.
