AI Calling for Real Estate Leads in India (2026): Site Visits Booked Minutes After the Enquiry, With the ₹ Cost and TRAI Rules

• ₹6per call minute on Adya by Amomic-AI, telephony included, price list checked November 2, 2026
• ₹13,999a month for 1,000 leads called for 1.5 minutes each on Growth, before GST — about ₹14 a lead
• 37%of 2,241 audited companies replied to a web lead within an hour; 23% never did (HBR, 2011, as reported in published summaries)
• 11+languages one agent calls in: English, Hindi and nine more Indian languages

AI calling for real estate leads means an agent that phones each new enquiry from 99acres, Magicbricks, Housing.com, your website or a Meta ad within minutes during calling hours, introduces itself as a virtual assistant calling for your project, qualifies configuration, budget, location and timeline in the buyer’s language, and books a site visit into your team’s calendar. The transcript, the structured answers and the booking land in your sales inbox; a buyer who asks for a person is handed to one.

Adya by Amomic-AI is one agent for phone calls, WhatsApp, website chat and email, so the same agent that called the buyer sends the WhatsApp confirmation with the location pin and answers the follow-up question about possession that arrives at 10 pm. On our price list a 1.5-minute qualification call costs ₹9 in minutes; 1,000 leads a month on the Growth plan comes to ₹13,999 before GST. The sections below show the arithmetic, the script, and the regulations.

The volume is the first reason. Magicbricks said in a press release dated July 1, 2026 that it had crossed 8 million monthly unique visitors in comScore’s May 2026 data, and put 99acres at 7.6 million, NoBroker at 6.1 million and Housing.com at 5.6 million. Those are company-issued figures and they are visitors, not leads, but they describe the funnel every developer and channel partner draws from: a very large pool of buyers who send the same enquiry to three or four listings in one sitting.

The second reason is the clock. Portal leads arrive in bursts: when a campaign goes live, after 9 pm, on Sundays. A sales team of two works the list from 10 am the next day, in order, between site visits. The lead that landed at 9:40 pm on Saturday is called at 11:30 am on Monday, roughly 38 hours later, by which time the buyer has spoken to two other projects and may have booked one of them.

That audit is fifteen years old and American, and we cite it because it measured the queue, not because it measured Indian real estate. The mechanism is the same: a lead sits until a person is free. An AI agent removes the queue during calling hours, because it places the call when the lead arrives rather than when the roster allows.

An outbound call to a lead is a short, structured conversation with a fixed opening and a clear goal. On Adya the agent speaks the moment the call connects, and the opening is a fixed template rather than your inbound greeting: “Hello, this is {agent name}, a virtual assistant calling on behalf of {company}. Is now a good time to talk?” In Hindi + English mode the same line is “Namaste, main {agent name} bol rahi hoon —{company} ki virtual assistant. Kya abhi baat karne ka sahi samay hai?” It names the business, discloses that the caller is a virtual assistant before asking anything, and asks whether now suits; someone in a meeting says no and the call ends there. Details are in the outbound calls guide.

Then the errand, from the lead’s own row

Every call carries a purpose: “You enquired about a 2 BHK at Greenfield Heights on 99acres this morning; I’m calling to answer questions and, if it suits, book a site visit.” In a call batch the purpose and script are filled in per person from your CSV, so {{name}}, {{csv.project}}, {{csv.configuration}} and {{csv.source}} become the buyer’s name, the project they asked about, the configuration and the portal. A blank cell does not stop the call; the agent is told the value is not on file and never guesses it.

Qualification, in the buyer’s words

The agent works through three to five things to find out — configuration, budget band, preferred location, timeline, home-loan need — asking one at a time, skipping what the buyer has already volunteered, and moving on if they decline to answer. After the call the transcript is read against your form and only what the buyer actually said is recorded; a budget they did not mention stays empty rather than being inferred.

Site-visit booking into a real calendar

If the buyer is interested, the agent offers open slots from a booking calendar: your site office’s opening days and hours, visit length, a gap after each visit, how many visits may overlap, minimum notice and closed dates. The console and the agent book against the same rules, so they can never offer different times. A connected Google Calendar blocks slots where a timed event is marked Busy, and the booking is mirrored back. See Calendar and appointments and Google Calendar.

The WhatsApp confirmation and the same agent afterwards

A booked visit is confirmed on WhatsApp with the date, time and location, from an approved utility template that Meta bills to your WhatsApp Business account at ₹0.1150 on its India card effective October 1, 2026. Because Adya is one agent across channels, the buyer’s reply — “Can we make it 4 pm?” or “Possession kab hai?” — is answered by the same agent with the same knowledge base, and the reschedule lands in the same calendar.

On Adya a call batch has a Why is it calling? field of up to 600 characters and an optional Call script of up to 3,000 characters. The script is a guide, not lines to read out: the agent covers your points in your order, in its own words. Write steps, not paragraphs. Here is one that works for a residential project.

Every call then ends in one group: Completed, Partly done, Asked to call back, Not interested, Wrong person, Hung up early, No conversation or Didn’t pick up. The success rate at the top of the Outcomes tab is the share of people who picked up and said yes to a visit, and every field gets a chart as the batch runs. Export is a PDF report or a CSV with every person and every answer. The field-by-field reference is in Call outcomes and the outcome dashboard, and the variable syntax in Call scripts and variables.

Two ways to start the call

Both routes resolve the agent from the number you call out on: every number has exactly one agent deployed to it, so the buyer always hears the same agent for that caller ID. Campaign calls are paced — Gradual starts slowly and builds up — and held to calling hours of 10:00–19:00 in your workspace time zone, a rule that cannot be loosened. The review step warns when a list will not finish today, for example Only 73 of these 1,318 calls will be placed today.

Retries, without harassment

Try again if nobody answers is off by default. Turned on, a number that rings out is called once more about two hours later, or the next morning if that would fall outside calling hours; nobody is called more than twice, and the second call carries the same brief. A separate switch, Call back when a call ends halfway, calls a buyer whose call dropped with answers missing once more, and that call knows what they already said. Frequency caps hold one message per campaign per person per day and at most four campaign contacts per person per week across all campaigns. See Sending rules.

Handoff to your sales team

Three things reach your team. Hot leads: a Completed call with a visit booked appears in Calls and the Inbox with the AI summary, the structured answers and the full transcript, for your team to assign to the closer who will attend. Live transfer: a buyer who asks for a person during live hours is offered one; human handoff has a transfer-sensitivity slider from 0% to 100% in steps of 5, so a sales line can be set to hand over early on price negotiation or a request for a specific executive. Outside live hours the agent books the visit or creates a ticket with the number and reason. Setup is in the human handoff guide.

A project in Pune gets Marathi and Hinglish; one in Coimbatore gets Tamil; an NRI campaign gets English with a Punjabi parent on the line. Adya calls in 11+ languages — English, Hindi, Bengali, Marathi, Tamil, Telugu, Gujarati, Kannada, Malayalam, Punjabi and Odia — with one agent and one knowledge base for all of them.

Outbound calls open in English when the number’s opening language is English and in Hindi/English when it is Hindi; for other opening languages the opening follows the call mode. Two numbers in one workspace can run two modes, so a Hyderabad project line can be All Indian languages while the NRI line is English only. Test with a real buyer-style call: a Hinglish caller who quotes a budget in lakh, interrupts the slot list and asks “possession kab tak?”. See Languages.

Prices are Adya’s list checked November 2, 2026: voice ₹6 a minute with telephony included; Starter ₹2,999 a month plus ₹399 a month for an Indian number; Growth ₹7,999 a month with 500 call minutes and one number included; GST at 18% on top. Assume one connected qualification call of 1.5 minutes per lead. Unanswered attempts are not billed as talk time, and a buyer who says “not now” in 20 seconds costs ₹2. Adya's rate is one row of our November 2026 study of published AI calling rates in India; the per-lead sums below use ours.

One ceiling to plan around before the rupees: each workspace has a daily safety budget of 100 outbound calls, or 600 reserved call minutes, per UTC day. Every accepted call reserves its worst-case length before dialling, and with the standard 10-minute cap that means the minute budget is reached at about 60 calls a day. Past it the API returns 429 daily_call_budget and a campaign’s final check blocks or warns with voice_daily_cap. The 1,000-lead row fits comfortably: on an even flow that is about 35 first calls a day, plus a handful of retries and call-backs, with visit confirmations going out on WhatsApp rather than by phone — roughly 40 of the 60 calls the budget allows. The 3,000-lead row is about 100 first calls a day on a 30-day month, so it needs the budget raised with us first, which is an Enterprise conversation, as does any launch-day burst of more than about 60 calls.

The WhatsApp confirmation and one reminder per booked visit add 2 × ₹0.1150 = ₹0.23 per visit on Meta’s India card, so 200 booked visits a month are ₹46 in Meta fees. Conversations on the plan cover the WhatsApp and web-chat threads; beyond the plan quota (1,000 on Starter, 5,000 on Growth) a conversation is ₹2. If your average call runs 3 minutes rather than 1.5, the minutes line doubles: 1,000 leads become ₹7,999 + 2,500 × ₹6 = ₹22,999 on Growth before GST. Use your own call length from the first week’s transcripts.

The telecaller line

Indeed India’s salary page for “Caller” roles shows an average of ₹16,447 a month from 1.7k reported salaries, updated June 5, 2026, with city averages from ₹19,283 in Thane to ₹30,064 in Noida. Real estate telecaller listings we saw on job boards in November 2026 ran from ₹10,000 fixed for a fresher to ₹30,000 fixed plus closing incentives for one to four years’ experience in Gurugram; there is no single published figure for the role, so use your own offer letters. To that add employer PF and ESI where applicable, a seat, a SIM and a dialer licence, paid leave and the cost of the two weeks it takes a new hire to learn the project.

The comparison is not only rupees. A telecaller works one list in order and reaches the Saturday-night leads on Monday; the agent reaches them at 10:00 on Sunday if you allow Sunday calling. A telecaller’s notes are a line in a sheet; the agent’s are structured fields, a transcript and, if you turn on Record calls for the number, a recording with a spoken notice to the buyer. And your telecallers do not disappear: the first call is the repetitive part, and the visits, the negotiation and the follow-up with a hot buyer stay with people. See the full price list, and the voice pricing method in AI voice agent pricing in India.

This section states what the regulations say and links them; it is not legal advice. TRAI’s Telecom Commercial Communications Customer Preference Regulations, 2018 (TCCCPR) govern commercial calls and messages, and the Second Amendment Regulations of February 12, 2025 tightened them; the complaint-mechanism changes took effect on April 13, 2025. No TRAI regulation bans automated or AI voice calls. What is regulated is the commercial call itself: who sends it, from which number series, with what consent, and at what frequency of complaints the sender is cut off.

The practical consequences for a developer or channel partner: register as a sender on your operator’s DLT platform, take a 140-series number for promotional calls, record where each lead came from, and keep the opt-out working — on Adya anyone on a channel’s opt-out list is removed at launch and the rule is enforced on every campaign, with no toggle to turn it off. The full walk-through, including the 2025 penalty ladder, is in DLT and TRAI rules for AI calling in India.

RERA, 2016: the call is marketing

The Real Estate (Regulation and Development) Act, 2016 (Act 16 of 2016) received assent on March 25, 2016 and has been fully in force since May 1, 2017. Section 3(1) says no promoter shall advertise, market, book, sell or offer for sale any project that is not registered with the state authority. Section 11(2) requires every advertisement or prospectus to carry the project’s RERA registration number and the Authority’s website address prominently; several state authorities, such as West Bengal’s, have extended that to social media and WhatsApp promotion, with placement and font-size rules of their own. Section 12 makes the promoter liable for an incorrect or false statement in an advertisement or prospectus on which a buyer paid, and a buyer who withdraws because of it is returned their money with interest. Section 9 requires real estate agents to register and quote their registration number. Section 59(1) allows a penalty of up to 10% of the estimated project cost for marketing an unregistered project.

DPDP Act, 2023: budgets and phone numbers are personal data

The Digital Personal Data Protection Act, 2023 (Act 22 of 2023) passed Lok Sabha on August 7 and Rajya Sabha on August 9, 2023. Section 6(1) requires consent that is free, specific, informed, unconditional and unambiguous, given by a clear affirmative action; Section 5 requires a notice before or with the request for consent describing the data and the purpose. Section 8(7) requires erasure when the purpose is no longer served unless a law requires retention. The Schedule sets penalties of up to ₹250 crore for failing to take reasonable security safeguards. The DPDP Rules, 2025 were notified as G.S.R. 846(E) on November 13, 2025; Rule 3 (notice) and Rules 5 to 16 — consent mechanics, security, breach reporting within 72 hours to the Board, retention — come into force 18 months later, on May 13, 2027.

For a lead call that means: say what you collect and why in one sentence, collect only what qualification needs (no PAN, no Aadhaar, no salary slips over the phone), put the full notice in the WhatsApp confirmation and on the enquiry form, and set a retention period for leads that never booked. On Adya each contact carries per-channel permission — allowed, opted out or not recorded — and a contact can be forgotten from the console. The checklist is in DPDP Act checklist for WhatsApp and AI chat.

Measure one number: site visits booked per 100 leads, this fortnight against last. Everything else is diagnostics.

Days 1–3 — facts and numbers. Put the registered project facts in the knowledge base: RERA number, configurations with carpet areas, price bands, possession date, site address and landmarks, visit timings. Take a 140-series number through your operator and complete DLT sender registration; on Adya add the number under Phone / Voice with KYC. Set the call language per number.
Days 4–5 — calendar and WhatsApp. Create the site-visit booking calendar: days, hours, 45-minute visits, a 15-minute gap, up to three visits at once if you have three executives. Connect Google Calendar and test both directions. Submit the visit-confirmation and reminder templates as utility with the RERA number in the footer.
Days 6–7 — script and outcome form. Write the purpose and the eight-step script above with your variables, build the Outcome form from the Qualify leads template with Interested in a site visit as the success rule, and read the review step’s “For the first recipient” view to catch a variable pointing at the wrong column. Make the Hinglish test call.
Days 8–12 — go live on one portal. Route one portal’s leads through the API callback during 10:00–19:00. Because API calls are not DND-scrubbed or hour-limited by the platform, your integration scrubs each number against the DND registry and records the consent source (portal, form, date) before it dials. Then run one call batch on the backlog from the last 30 days with Gradual pacing and Try again if nobody answers switched on, keeping the day’s total inside the daily call budget. Read every transcript each morning; fix the knowledge base, not the prompt, when it gets a fact wrong.
Days 13–14 — measure and decide. Visits booked per 100 leads, show-up rate against the previous fortnight, average call length (which sets your minutes bill), and the Not interested share, which tells you about the portal more than about the agent.