The Real Estate Voice AI Playbook: From Enquiry to Appointment
A practical guide to designing real estate voice AI playbooks that classify enquiries, capture useful context, book appointments, and create clear next actions.
A real estate voice agent should not just answer the phone.
It should understand the enquiry, ask the right questions, capture useful context, and move the customer towards the right next step.
That sounds simple. In practice, it is where a lot of voice AI projects become weak.
The agent sounds polite, but the conversation does not go anywhere. Or it asks too many questions. Or it pushes for a booking too early. Or it treats every caller like the same type of lead.
Real estate does not work like that.
A seller enquiry is different from a buyer enquiry. A rental enquiry is different from an appraisal request. A missed call is different from an inbound call. A past database contact is different from someone calling about a property they saw five minutes ago.
A good real estate voice agent needs a playbook.
A playbook is not a script. It is a clear operating model for how the agent should handle a specific customer journey.
Start with the customer journey, not the agent script
The first mistake is starting with:
"What should the agent say?"
The better starting point is:
"What is the customer trying to do?"
In real estate, the customer may be trying to:
- book an appraisal
- ask about a listing
- inspect a property
- understand whether a property is still available
- speak to a property manager
- follow up after an open home
- ask about selling in the future
- reconnect after an old enquiry
- find out who they should speak to
These are different journeys.
The agent should not flatten them into one generic conversation.
A useful playbook starts by identifying the customer's intent.
The first job is intent detection
Intent detection does not need to sound technical.
It can be as simple as asking:
"What were you calling about?"
or:
"Are you calling about a property, selling your own place, or something else?"
The answer tells the agent which path to follow.
For example:
Customer says:
"I wanted to know if the apartment on Victoria Road is still open this weekend."
Likely intent:
Buyer listing enquiry
Better next step:
Identify the property, capture contact details, help with inspection information or team follow-up.
Different example:
Customer says:
"We are thinking of selling later this year and wanted to know what our house might be worth."
Likely intent:
Seller appraisal enquiry
Better next step:
Capture property address, selling timeline, motivation if naturally shared, and help book an appraisal.
The agent should not treat these as the same call.
Seller enquiries need a careful path
Seller enquiries are valuable, but they are also sensitive.
A potential seller may be early in their thinking. They may not want to feel pushed. They may not know their timeline. They may want a price estimate, but the agent should not pretend to provide one.
A good seller enquiry path should capture:
- customer name
- phone number
- property address
- suburb
- selling timeline
- reason for selling, if naturally shared
- whether they want an appraisal
- preferred follow-up time
- preferred contact method
But it should not feel like an interrogation.
A weak agent might say:
"What is your name? What is your email? What is your phone number? What is your property address? When are you selling?"
A better agent adapts to the conversation:
"That makes sense. I can help organise a proper appraisal. What is the property address?"
Then:
"Are you looking to sell soon, or are you still working out timing?"
Then:
"Would you prefer the team to call you today or tomorrow?"
Short. Useful. Natural.
Buyer enquiries need property context
Buyer enquiries are often more specific.
The customer may be asking about a property they saw online, an inspection time, a price guide, or whether the property is still available.
The first challenge is identifying the property.
A good buyer enquiry path should capture:
- property of interest
- inspection preference
- buyer name
- phone number
- email if needed
- buying status
- urgency
- whether they want similar listings
- whether they need a callback
For example:
"Which property were you looking at?"
If they do not remember:
"No problem. Do you remember the suburb, street, or anything about the listing?"
That feels better than forcing the customer into a rigid form.
Once the property is identified, the agent can help with the next step.
If inspection information is available:
Share or confirm inspection details.
If calendar booking is available:
Offer available inspection slots.
If unsure:
Create a clear follow-up task for the listing agent.
Rental enquiries should not be forced into sales flows
Rental enquiries are common, repetitive, and often time-sensitive.
They should not be handled like seller or buyer leads.
A rental enquiry may involve:
- inspection times
- application process
- property availability
- lease start date
- pets
- parking
- maintenance
- property manager contact
- urgent tenant issues
A good agent needs to classify the difference between a prospective tenant and an existing tenant.
For example:
Prospective tenant:
"I saw a rental listing and wanted to inspect it."
Existing tenant:
"My hot water is not working."
These are not the same workflow.
The first might need inspection information.
The second might need urgent property management escalation.
A real estate playbook should keep those paths separate.
Missed calls need a different tone
A missed-call follow-up is not the same as an inbound call.
On an inbound call, the customer has chosen to call now.
On a missed-call callback, the agent is interrupting them later.
The opening should respect that.
Good missed-call opening:
"Hi, this is Chloe, an AI assistant calling back on behalf of Greenfield Realty. It looks like we missed your call earlier. Is now still a good time?"
That is much better than:
"Hi, would you like to book an appraisal?"
Why?
Because the agent does not know why the customer called yet.
A missed-call playbook should first understand intent, then choose the right path.
Missed call path:
- explain why the agent is calling
- ask if now is still a good time
- ask what the customer was calling about
- classify the intent
- capture only the details needed for that intent
- create the right next action
Inbound calls can be more direct
Inbound calls can start differently.
The customer called the business. They expect help.
A good inbound opening might be:
"Thanks for calling. I’m Chloe, the AI assistant for the team. What can I help you with today?"
That is enough.
No need for a long explanation. No need to over-apologise. No need to ask for too much upfront.
The agent should listen first.
Then it should route the conversation.
Do not ask every question on every call
This is one of the most important design rules.
A voice agent should not ask every possible question just because the field exists in the CRM.
A seller enquiry may need property address and timeline.
A buyer enquiry may need property of interest and inspection preference.
A rental enquiry may need lease timing and property details.
A missed-call enquiry may only need reason for call and callback preference.
Good playbooks use conditional capture.
That means the agent captures what matters for the path the customer is actually on.
For example:
If seller enquiry:
Capture property address, timeline, motivation, appraisal preference.
If buyer enquiry:
Capture property of interest, inspection preference, buying status.
If rental enquiry:
Capture property, inspection/application question, urgency.
If existing client:
Capture issue type, urgency, and preferred callback.
This keeps the conversation shorter and more respectful.
The agent should know what not to say
Real estate agents deal with questions that require care.
Customers may ask:
"What do you think my property is worth?"
or:
"Can you guarantee it will sell above this price?"
or:
"Should I accept an offer?"
or:
"Can I break my lease?"
An AI agent should not improvise on those topics.
It can still be helpful.
For example:
"I cannot give a reliable price estimate on this call, but I can help organise an appraisal so the team can review the property properly."
That is a good answer.
It sets a boundary and creates a next step.
A playbook should clearly define:
Agent can:
- capture enquiry details
- explain process at a high level
- book or request follow-up
- confirm known appointment or inspection information
- route the customer to the right team
Agent cannot:
- provide legal advice
- guarantee a sale price
- negotiate commission
- promise an outcome
- invent property or market details
Realtime actions should be limited and useful
Some actions make sense during the call.
For example:
- checking appraisal availability
- booking a consultation
- confirming inspection times
- identifying a known contact
- routing urgent property management issues
Other actions can happen after the call.
For example:
- updating the CRM
- tagging lead type
- attaching the summary
- creating a follow-up task
- notifying the team
- adding the customer to a nurture sequence
The agent should not slow the call down by doing too much live.
A useful design rule is:
Do live only what the customer needs during the call. Do the rest after the call.
That keeps the conversation moving.
The playbook should produce structured output
At the end of the call, the team needs more than a transcript.
They need a clear operational result.
For a seller enquiry:
Call type: Seller enquiry
Customer intent: Wants appraisal
Property suburb: Castle Hill
Selling timeline: 3 to 6 months
Motivation: Upsizing
Lead temperature: Warm
Appointment status: Preferred time captured
Recommended next action: Senior agent callback today
Review required: No
For a buyer enquiry:
Call type: Buyer listing enquiry
Property of interest: 8 Sample Street, Parramatta
Question: Inspection availability
Buyer status: Actively looking
Preferred follow-up: SMS
Recommended next action: Send inspection details and notify listing agent
Review required: No
For a property management issue:
Call type: Existing tenant issue
Issue: Hot water not working
Urgency: High
Recommended next action: Escalate to property management
Review required: Yes
Structured output is what turns the call into a workflow.
The team should see the next action, not just the summary
A common mistake is producing a nice summary and stopping there.
A summary is helpful, but the team still needs to decide what to do.
A stronger system gives the next action.
For example:
Weak:
"Customer is interested in selling."
Better:
"Warm seller lead. Call today after 4 pm to discuss appraisal. Customer is thinking of selling in 3 to 6 months and is upsizing."
The second version is easier to act on.
The goal is not just documentation.
The goal is momentum.
Human review should be part of the flow
Not every conversation should trigger automation immediately.
Some calls should go to review.
For example:
- customer is angry
- customer asks for legal advice
- customer asks for a guaranteed price
- customer gives sensitive personal context
- customer asks for a human
- agent confidence is low
- transcript and structured output do not match cleanly
A good playbook should define review triggers.
Send to review when:
- the customer asks for advice outside scope
- the customer is upset
- the agent is unsure of the outcome
- the customer gives conflicting information
- the next action could affect trust or compliance
This makes the agent safer and easier to trust.
A real estate playbook should be visible to the business
The business should not have to guess how the agent behaves.
They should be able to see:
- what journeys the agent handles
- what questions it asks
- what fields it captures
- what outcomes it can assign
- what actions happen in real time
- what actions happen after the call
- what triggers human review
- what the agent must not say
If those rules are hidden inside a long prompt, the business cannot properly review them.
That is why playbooks should be visual and structured.
A prompt can make the agent sound good.
A playbook helps the team understand and trust how it behaves.
Example: seller appraisal playbook
Here is a simple seller appraisal playbook.
Use case:
Seller appraisal enquiry
Goal:
Understand whether the caller is thinking of selling and help book an appraisal or create a follow-up task.
Conversation path:
- understand intent
- capture property address
- capture selling timeline
- ask about motivation if naturally relevant
- offer appraisal booking or callback
- confirm next step
Capture:
- customer name
- phone number
- property address
- suburb
- selling timeline
- reason for selling
- preferred appointment time
- preferred follow-up method
Agent should:
- ask one question at a time
- keep responses short
- avoid pressure
- avoid price estimates
- explain that the team will provide a proper market view
Realtime actions:
- check calendar availability if booking is enabled
- book appraisal if customer agrees
Post-call actions:
- update CRM
- attach call summary
- tag as seller lead
- create follow-up task if not booked
- notify assigned agent
Escalate when:
- customer asks for guaranteed sale price
- customer asks for legal advice
- customer is upset
- customer asks for a human
That is much stronger than:
"You are a helpful real estate assistant."
Example: buyer listing enquiry playbook
Buyer enquiries need a different path.
Use case:
Buyer listing enquiry
Goal:
Help the caller with a specific property enquiry and capture enough context for the listing agent or follow-up workflow.
Conversation path:
- identify property
- understand question
- capture buyer contact details
- ask about inspection interest
- provide known next step or create follow-up task
Capture:
- customer name
- phone number
- property of interest
- inspection preference
- buying status
- preferred follow-up method
Agent should:
- avoid inventing property details
- not guess availability
- not negotiate price
- route listing-specific questions when needed
Realtime actions:
- confirm inspection availability if connected
- send or request inspection details
Post-call actions:
- update CRM
- tag as buyer enquiry
- attach property of interest
- notify listing agent if high intent
The structure is similar to the seller playbook.
The behaviour is different.
Example: rental enquiry playbook
Rental enquiries need their own path too.
Use case:
Rental enquiry
Goal:
Understand whether the caller is a prospective tenant, existing tenant, or landlord and route them correctly.
Conversation path:
- classify caller type
- identify property or issue
- capture urgency
- provide next step or route to property management
Capture:
- caller name
- phone number
- property address if relevant
- enquiry type
- urgency
- preferred follow-up
Agent should:
- separate prospective tenant enquiries from existing tenant issues
- escalate urgent maintenance
- avoid legal or lease advice
- create a clear handoff if property management is needed
Post-call actions:
- update CRM or property management system
- create task
- notify property management if urgent
This avoids one of the biggest problems in automation: forcing every caller into the same sales process.
What good feels like for the customer
A good real estate voice agent should make the customer feel:
- the agency responded quickly
- the agent understood why they called
- the questions were relevant
- the call did not feel like a form
- the agent did not overpromise
- the next step was clear
- the team will know what happened
That is the bar.
Not sounding clever.
Not having a long prompt.
Not collecting every possible detail.
Just helping the customer move forward.
What good feels like for the team
For the team, a good playbook should reduce ambiguity.
They should not have to ask:
"What did the customer want?"
or:
"Who needs to call them?"
or:
"Was this a seller, buyer, tenant, or something else?"
They should see:
Lead type: Seller appraisal
Customer intent: Wants market view
Timeline: 3 to 6 months
Temperature: Warm
Next action: Call today after 4 pm
Review required: No
That is what turns a conversation into something useful.
How DialoGrove thinks about real estate voice agents
At DialoGrove, we do not think real estate teams need a blank prompt box.
They need a way to turn real lead journeys into clear, reviewable agent behaviour.
That means:
- scenario paths
- captured fields
- outcomes
- guardrails
- realtime actions
- post-call workflows
- review triggers
- CRM-ready outputs
Real estate is a strong first use case because the problem is visible.
Calls matter. Timing matters. Lead context matters. Follow-up matters.
But the same principle applies to any business where customer conversations need to become structured next actions.
A good voice agent does not just answer.
For the seller-specific strategy beyond the playbook template, see our seller lead qualification guide.
For the real-estate-specific strategy behind dormant database outreach, see our past buyer reactivation guide.
It understands the journey, captures what matters, respects the boundary, and helps the team act.