IRDAI-Compliant AI Calling Bot for Insurance Sales, Renewals & Cross-Sell: The India Playbook

Every insurance company in India wants to run AI outbound calling. Most of them stall at the same question: "Is it IRDAI compliant?"

The answer is yes — but only if you build it right. The Insurance Regulatory and Development Authority of India has clear requirements around what must happen on any automated call that touches a policyholder or an insurance prospect. Get those requirements wrong and you're not just looking at a regulator fine; you're looking at call blocking, policyholder complaints, and a compliance audit.

Get them right, and you unlock the most powerful lead conversion and retention tool available to insurance companies in India today.

This guide is for insurance companies, bancassurance teams, intermediaries, and brokers who want to deploy an AI calling bot for insurance sales and renewals — and need to know exactly how to do it without running into IRDAI, TRAI, or DPDP Act 2023 issues.

What IRDAI Actually Says About Automated Insurance Calls

The regulatory framework for automated outbound calling in insurance comes from three overlapping sources.

IRDAI's Guidelines on Outsourcing of Activities (2017, updated 2023) require that any automated communication with policyholders must: (a) identify the insurer by name at the start of the interaction, (b) clearly state the purpose of the call before any data collection begins, and (c) provide a clear mechanism for the customer to opt out of future automated calls. These requirements apply equally to AI voice agents and to IVR systems.

TRAI's Telecom Commercial Communications Customer Preference Regulations (TCCCPR 2018) govern the scrubbing of numbers against the National Do Not Disturb (NDND) registry before any commercial communication. Insurance calls fall under "Financial Products and Services" — a regulated category — which means you must scrub every number against the NDND registry before dialling, and you must be registered as a Principal Entity with the relevant telecom operator. DND violations carry fines of Rs 25,000 per complaint.

DPDP Act 2023 adds a consent layer on top of IRDAI and TRAI. Any AI call that collects or processes personal data — which includes capturing the policyholder's spoken confirmation of a premium amount, their stated health condition, or their bank account preference — requires documented consent. The consent must be purpose-specific and must be stored against the individual record.

The practical upshot: an IRDAI-compliant AI calling bot for insurance is not difficult to build, but it requires the right disclosure sequence, a consent architecture that creates an auditable log, and NDND scrubbing before every dial.

The 5 Mandatory Disclosures Every AI Insurance Call Must Include

Before the AI says anything about a product, renewal, or cross-sell offer, five disclosures must happen — and they must happen in sequence.

Disclosure 1: Identity of the insurer. "This is an automated call from [Insurer Name], a company registered with IRDAI." The IRDAI registration number is optional but builds trust. Never lead with the intermediary or distributor name without mentioning the underlying insurer.

Disclosure 2: Purpose of the call. "This call is regarding your [policy type] policy number [last 4 digits]." For sales calls to prospects: "This call is regarding a [term/health/motor] insurance plan that you enquired about on [date/channel]."

Disclosure 3: Automated call declaration. "You are speaking with an AI voice assistant. If you would like to speak with a human agent, say 'connect me to an agent' or press 0 at any time." This is both an IRDAI requirement and a basic consumer protection standard.

Disclosure 4: Opt-out mechanism. "To stop receiving these calls, say 'do not call' or press 9." The opt-out must be honoured immediately and must suppress the number from future calling lists within 24 hours. TRAI requires this for all commercial communications; IRDAI reinforces it for insurance.

Disclosure 5: Recording disclosure. "This call may be recorded for quality and compliance purposes." Required under IRDAI's call centre guidelines and good practice under DPDP Act 2023.

These five disclosures take approximately 25–35 seconds. Do not try to compress them. Regulators look for complete disclosures; policyholders who experience omissions are more likely to complain.

Consent Architecture for AI Insurance Calls

The DPDP Act 2023 requires that consent for processing personal data be freely given, specific, informed, and unambiguous. For an AI insurance calling program, this means three tiers of consent, each logged differently.

Tier 1: Pre-existing consent from the policy application. When a customer took out a policy, they typically consented to receive communications from the insurer about their policy. This consent covers renewal reminders and premium payment reminders. It does not automatically cover cross-sell calls or new product offers unless the application form specifically included that language.

Tier 2: Fresh consent for outbound sales calls to prospects. If you're calling a prospect from a lead aggregator list, their web form submission, or a bancassurance referral, you need documented consent that specifically covers AI voice outreach for insurance products. The lead source must capture and log this consent. Your AI calling platform should verify consent status before dialling.

Tier 3: In-call consent for data capture. Any time the AI captures new personal data during the call — health conditions, nominee details, income bracket, preferred payment method — it should announce: "I'm going to note your response. Do you confirm this is accurate and that we may use it to process your request?" This creates an in-call consent event that should be logged with a timestamp against the customer record.

Most India-first voice AI platforms support consent logging natively. If yours doesn't, that's a red flag.

Use Case 1: Policy Renewal Reminders — Compliant Script Architecture

Policy renewal is where AI calling delivers the fastest, most measurable ROI for insurers. The process is straightforward: the policy has an expiry date, renewal is due, the AI calls to prompt action. IRDAI compliance here is relatively low friction because the call is about an existing policy that the customer already holds.

A compliant renewal reminder call follows this arc:

Opening + disclosures (30 seconds): "Hello, this is Priya, an AI assistant from [Insurer Name], registered with IRDAI. I'm calling about your [policy type] policy ending [last 4 digits] which is due for renewal on [date]. This is an automated call. You can say 'connect to agent' at any time to speak with a human. To stop receiving these calls, say 'do not call'."

Confirmation of contact (10 seconds): "Am I speaking with [customer name]?" Wait for confirmation before proceeding with any policy details. This protects against discussing policy details with the wrong person — a data privacy requirement.

Renewal prompt (45 seconds): "Your premium for the coming year is Rs [amount]. Would you like to renew your policy today? I can send you a payment link on your registered mobile number right now." If yes, generate the link and read out the UPI ID. If the customer asks for more time, offer a specific callback date.

Cross-sell opportunity (optional, 30 seconds): Only introduce cross-sell after the renewal is handled. "Since you're renewing your health cover, would you like to know about our top-up cover option that increases your sum insured for an additional Rs [amount] per month?" This sequencing matters — IRDAI disfavours calls that lead with cross-sell before handling the primary policy purpose.

Close + opt-out reminder: "Thank you, [name]. A summary will be sent to your registered email. As a reminder, to stop these calls, say 'do not call' or press 9."

Insurers using this script architecture report renewal completion rates of 18–28% on the first AI call — versus 8–12% for SMS-only reminders. Lapse rates on policies that receive AI renewal calls drop to single digits in most cohorts.

Use Case 2: Insurance Sales Calls — What AI Can and Cannot Say

Sales calls to prospects are more tightly regulated than renewal calls to existing policyholders. The key IRDAI constraint is that AI cannot give personalised financial advice — it can present product features and invite the prospect to speak with a licensed agent for advice.

What AI can do on a sales call:

• State the features of the product (sum insured, exclusions, premium range)
• Quote the premium for the prospect's stated age and coverage amount
• Schedule a callback with a licensed agent for the recommendation step
• Capture expressed interest, nominee preference, and coverage amount preference
• Send a product brochure or quote via SMS/WhatsApp post-call

What AI cannot do on a sales call:

• Recommend a specific policy as the "right" policy for the customer (this is financial advice requiring a licensed agent)
• Compare the insurer's product against a competitor's product by name
• Make guarantees about claim settlement ratios or future bonuses
• Collect payment details — premium collection for a new policy requires human agent involvement in most insurer workflows

The practical model that works: AI handles the top of the funnel (lead qualification, interest capture, appointment booking for a licensed agent), and human agents handle the advice-and-sale step. This division of labour respects IRDAI's agent licensing requirements while dramatically increasing the number of qualified prospects the human agents see each day.

Insurers using this AI-to-human handoff model report 3–5× increases in agent productivity — measured as qualified leads per agent per day — because the AI pre-qualifies interest, captures basic details, and schedules the agent callback at a specific time.

Use Case 3: Cross-Sell and Add-On Coverage to Existing Policyholders

Cross-selling to an existing policyholder is legally simpler than selling to a new prospect because the consent relationship already exists. The AI can introduce a relevant add-on — top-up health cover, a critical illness rider, a motor add-on — because it is a communication about the customer's existing insurance relationship.

The highest-ROI cross-sell calls follow a specific timing pattern:

• 30 days after policy issuance (while the customer is still engaged with the brand)
• At renewal time (capture additional coverage while the payment intent is active)
• After a claim event (customers who have recently experienced a claim are 2–4× more likely to add coverage)

For health insurance customers, the top-up cover cross-sell is the most consistently successful. "Your current sum insured is Rs 5 lakhs. Medical inflation means a 5-day hospitalisation in a metro hospital now averages Rs 4.2 lakhs. Would you like to know about a top-up cover that increases your sum insured to Rs 20 lakhs for Rs 480 per month?" converts at 12–18% in tested deployments.

The AI should never pressure a customer who declines. A single "Are you sure? This offer is only available until [date]" is acceptable. More than one follow-up within the same call, or a repeat call within 48 hours, approaches aggressive selling patterns that IRDAI has flagged in past circulars.

Use Case 4: Premium Payment Reminders with UPI Link Delivery

Premium payment reminders are high-frequency, high-compliance-sensitivity calls. They must be IRDAI-aligned, DPDP-compliant, and DND-scrubbed. They must not cause distress to the policyholder. And they are extremely effective — AI premium reminder calls drive payment completion rates of 35–50% on the first call among policyholders who have received the call.

The critical technical requirement here is post-call action. An AI premium reminder call that tells the customer "your premium is due" but can't send the UPI payment link immediately converts poorly. The most effective flows:

1. AI call confirms the policyholder is present
2. AI states the premium amount and due date
3. AI asks: "Can I send you a payment link on your registered mobile number right now?"
4. On confirmation, the payment link is delivered via SMS within 30 seconds
5. AI reads out the UPI ID and the amount
6. If the customer wants to pay by card/net banking, AI routes to a human payment agent or sends a URL

The UPI link delivery is where most voice AI deployments fail. Ensure your platform can trigger an outbound SMS within 30 seconds of the in-call confirmation event — not after the call ends, not via a manual process, but as a real-time API call during the call.

Insurers using this end-to-end flow see 35–50% same-day payment completion on reminder calls versus 10–15% for reminder-only calls without an immediate payment link.

DPDP Act 2023: The New Compliance Layer Every Insurance AI Program Needs

The Digital Personal Data Protection Act 2023 (effective 2024–2025 for most enterprises) adds requirements on top of IRDAI and TRAI that insurance AI calling programs must now account for.

Data minimisation. The AI should only collect data necessary for the specific call purpose. A renewal reminder call should not capture health status updates. A premium reminder call should not ask about income changes. The AI script should be bounded to what the call purpose requires.

Data residency. Call recordings, transcripts, and captured customer data must be stored in India. Verify that your voice AI platform's data centres are in India — not Singapore, Ireland, or US. DPDP Act requires Indian data residency for sensitive personal data, and insurance data qualifies.

Purpose limitation. Consent captured for a renewal reminder cannot be repurposed for a sales call to a new product without fresh consent. If your CRM uses a single consent record for all outbound communication, that model will not survive DPDP scrutiny. Insurance companies need campaign-level consent logging.

Right to erasure. Policyholders can request deletion of their call recordings and AI interaction logs. Your vendor must support a data deletion API that you can trigger from your CRM.

Most India-first voice AI vendors have built DPDP Act compliance features — data residency, purpose-bound consent, deletion APIs — into their 2025 and 2026 platform releases. Global voice AI platforms, particularly those built in the US or EU, often require significant customisation to meet these requirements.

Choosing an IRDAI-Aware Voice AI Vendor for Insurance

Not all voice AI vendors understand the insurance regulatory environment. When evaluating vendors, these are the non-negotiable requirements:

1. NDND scrubbing. Automated DND registry scrub before every dial. This must be a platform-level guarantee, not a manual export-import process.

2. Disclosure script enforcement. The platform must support mandatory script segments — disclosures that cannot be skipped or shortened, even if the agent reprograms the script. Compliance requires that the five disclosures always run before any product discussion.

3. Opt-out processing. In-call opt-out commands ("do not call", "remove me from your list") must trigger immediate suppression — not end-of-day batch processing. Real-time suppression is a TRAI requirement.

4. Consent logging. Every call should produce a consent log entry with timestamp, phone number, call recording ID, and the specific consent events that occurred. This log must be exportable for IRDAI audits.

5. India data residency. Confirmed on-paper, not just stated on a website. Ask for the data processing agreement (DPA) that specifies where data is stored and processed.

6. IRDAI disclosure templates. The vendor should have pre-built, legal-reviewed disclosure templates for insurance use cases — not generic voice scripts that you have to modify yourself.

7. Human escalation. Every AI call must have a working escalation path to a licensed human agent. "Press 0 for agent" is not enough — the escalation must be instant (under 5 seconds) and must pass the full call transcript to the human agent before they pick up.

Vendors who have deployed for insurers — life, health, motor, and general — understand IRDAI's nuances in ways that general-purpose platforms don't. Ask your vendor for their insurance customer list and for a reference call with an existing IRDAI-regulated client.

Benchmarks: What to Expect from IRDAI-Compliant AI Calling for Insurance

Indian insurers who have deployed IRDAI-compliant AI calling programs at scale report these benchmarks:

Policy renewal:

• Renewal completion rate on AI calls: 18–28% (vs 8–12% for SMS alone)
• Lapse rate reduction: 30–50% on called cohorts
• Cost per renewal via AI: Rs 40–80 (vs Rs 200–400 via human agent)

Premium payment reminders:

• Same-day payment completion: 35–50% on first reminder call
• Days-sales-outstanding reduction: 15–25 days improvement
• Cost per collection via AI: Rs 8–15 (vs Rs 60–120 via human collector)

Lead qualification for sales:

• Qualified leads per AI call campaign: 8–14% of called base express interest
• Agent productivity increase: 3–5× qualified leads per agent per day
• Lead-to-policy conversion: similar to human-qualified leads when AI quality is high

Cross-sell:

• Conversion on top-up health offers to existing policyholders: 12–18%
• Average premium uplift per cross-sell: Rs 3,000–8,000 per annum
• Best timing: 30-day post-issuance calls and renewal-time offers

These numbers are achievable with a well-built IRDAI-compliant AI program. The gap between "we ran a pilot and got 3% conversion" and "we're running at 22% renewal completion" is almost always the script quality, the IRDAI disclosure architecture, and the post-call action (payment link, agent callback).