Rental brokers in Gujarat lose most of their leads in the first hour. Someone clicks an ad for a 3 BHK in a good area, fills the form, and calls the next three brokers too. Whoever calls back first, in Gujarati, and books a visit usually gets the deal.
A Gujarati AI caller can make that first call within minutes, any time of day. It asks what the tenant needs, checks budget and timeline, offers a site-visit slot and grades the lead hot, warm or cold, so your team spends its time on visits, not on calling forms back.
This guide uses the call design we built for a Gujarat rental brokerage: three agents for three kinds of rental lead, the questions each one asks, the Gujarati lines they use, and the guardrails that keep them honest.
Try it yourself: Build this agent free on Edesy — self-serve, no demo needed.
Why do rental leads need their own call?
Rental leads aren't buyers. They move faster, decide on smaller details, and come in three very different kinds:
- Home seekers want a 2, 3 or 4 BHK or a bungalow, in a specific area, within a budget, often within weeks.
- Business tenants want a shop, showroom or office, and care about carpet area and location.
- Landlords want to rent out a property to a good tenant and want to know you'll send serious people.
One script for all three sounds wrong to everyone. Build one agent per lead type, each triggered by the ad or form the lead came from.
The questions each agent asks
Ask one question at a time. If the lead already said something, the agent skips that question.
| Home seeker | Business tenant | Landlord |
|---|---|---|
| Property type (2/3/4 BHK, bungalow) | Space type (shop, showroom, office) | Property type and BHK |
| Preferred area | Preferred area | Area of the property |
| Monthly rent budget | Carpet area needed | Expected monthly rent |
| Move-in timeline | Monthly budget | Available from |
| Family, bachelor or company | Type of business | Preferred tenant (family, bachelor, company) |
| Furnished or unfurnished | When they need the space | Furnished or unfurnished |
| Site-visit slot | Site-visit slot | Verification or listing visit slot |
Before offering the visit, the agent reads the key facts back: "So a 3 BHK in your chosen area, around Rs 20,000, for a family, correct?" Gujarati speech recognition can mishear numbers, and the read-back catches it before the details reach your team.
What does the call sound like?
The agent speaks everyday Gujarati and keeps the English words people actually use. A few lines from the home-seeker agent:
- Opening: "નમસ્તે! હું અમારી real estate ટીમ તરફથી વાત કરું છું. તમે ભાડેથી ઘર શોધવા માટે અમારી જાહેરાત પર સંપર્ક કર્યો હતો. તમારી સાથે બે મિનિટ વાત કરી શકું?"
- Budget: "મહિને ભાડા માટે તમારું budget આશરે કેટલું છે?"
- Timeline: "ક્યારે shift થવાનું plan છે — તરત, એક મહિનામાં, કે પછી?"
- Visit: "તમારી પસંદ મુજબ verified options માટે હું direct site visit book કરી આપું — કયો દિવસ ને સમય અનુકૂળ રહેશે?"
Two things make it sound human rather than like a form:
- A short reaction before every question ("હા જી, સરસ.", "ઓહ, સમજી ગઈ."), with varied wording.
- Every turn ends with a question or a next step. A bare "okay" leaves the caller waiting, and the call stalls.
A tip from building these agents: write the instructions to the AI in English but every spoken line in Gujarati. The AI follows English instructions more reliably, and the Gujarati lines fix the exact register you want customers to hear.
Grading leads hot, warm and cold
Give the agent a written rule instead of letting it guess:
- Hot: clear requirement, a budget, wants a visit, needs the place within about a month.
- Warm: interested, but not urgent or still deciding.
- Cold: just browsing.
At the end of every call the agent saves the grade, all the answers and a one-line English summary. Send them to Google Sheets or your CRM, and your team calls hot leads first, the same day.
Guardrails that keep the agent honest
- Never quote rent or fees. If asked "what's your brokerage?", the agent says your team will send the details on WhatsApp. It never invents a number.
- Never invent availability. If unsure, it says the team will confirm.
- Voicemail: if a machine answers, it ends the call without leaving a pitch.
- Wrong number: it apologises in Gujarati and ends the call.
- Call limits: cap calls at about five minutes and end politely after 15 seconds of silence.
- "Where did you get my number?" It says the person responded to your ad, which is true.
What does it cost?
An example with assumptions to replace with your own:
| Assumption | Value |
|---|---|
| New rental enquiries per month | 500 |
| Average call length | 2 minutes |
| Rate | Rs 4 to Rs 6 per minute, depending on plan |
500 x 2 minutes = 1,000 minutes, roughly Rs 4,000 to Rs 6,000 a month, to call every enquiry within minutes, including evenings and Sundays. With bring-your-own AI keys, Edesy's platform fee is Rs 1.5 per minute.
How to set it up on Edesy
- Create three agents (home seeker, business tenant, landlord) with the questions above, in Gujarati.
- Add your listings and areas to each agent's knowledge base so it can mention the areas you cover. Keep prices out.
- Save the answers and grade at the end of each call, and send them to Google Sheets or your CRM.
- Trigger a call on every new enquiry from your ad forms, and run callback campaigns with calling windows and retries for leads who didn't pick up.
- Test with your own team first. Call each agent a few times in Gujarati, with different budgets and areas, and fix any question that sounds stiff.
Start free
Sign up and build your first Gujarati agent in minutes, with Rs 50 of free credit and no demo call needed.