Reviewing AI Output Before It Goes to a Client.

Before sending anything, confirm:

• names

• dates

• amounts

• property details

• rent figures

• quotes

• contractor advice

• lease dates

• actions already taken

• promised timeframes

AI can easily confuse details, particularly when it is working from long notes or email threads.

A professional-looking mistake is still a mistake.

AI often tries to make communication complete.

That can mean adding information that was never provided.

Watch for statements such as:

The contractor has confirmed the issue is now resolved.

when the contractor only confirmed they attended.

Or:

The landlord has approved the repair.

when approval is still outstanding.

If the source information does not support the statement, remove it.

AI can help explain a recommendation.

It should not quietly create one.

Check any wording such as:

• I recommend...

• We suggest...

• The best option is...

• You should...

• We believe...

Ask:

Did I actually decide this?

If not, either remove it or make the decision yourself first.

AI can sometimes produce communication that is technically correct but does not suit the relationship.

It may sound:

• too formal

• too soft

• too aggressive

• defensive

• overly apologetic

• unnecessarily wordy

• obviously automated

Read the message as the recipient would.

Would this sound appropriate coming from you?

Be particularly careful with phrases such as:

This will be resolved tomorrow.

The contractor will attend today.

The landlord will approve this.

We will have an answer by 3 pm.

If the outcome is not confirmed, the wording should not suggest that it is.

A safer message may be:

I will contact the contractor today and update you by 3 pm.

That promises an action the PM can control.

AI should not be trusted to determine the legal position without verification.

If the communication relates to:

• rent increases

• access

• repairs

• breaches

• notices

• arrears

• compensation

• termination

• tribunal matters

check the current requirements in the relevant state or territory before sending.

Pay particular attention if AI has included:

• notice periods

• deadlines

• legal rights

• prescribed processes

• statements about what a landlord or renter "must" do

Confirm them independently.

AI can sometimes include every detail it was given.

The recipient may not need all of it.

Before sending, remove:

• unnecessary internal notes

• comments about other clients

• staff discussions

• sensitive personal information

• information that should remain internal

• background that does not help the recipient

The final communication should contain what the client needs, not everything the AI knows.

AI may occasionally make a message more forceful than intended.

Watch for phrases such as:

You have failed to...

You are responsible for...

You must immediately...

This is unacceptable...

Sometimes strong language is appropriate.

Often, a factual version will achieve more.

For example:

We have not yet received approval for the repair and need a decision by Friday so the matter can progress.

That is clear without unnecessarily escalating the tone.

A well-written email can still be ineffective if nobody knows what to do next.

Before sending, ask:

• Does the recipient need to make a decision?

• Do they need to provide information?

• Are we simply updating them?

• When will the next update happen?

• Who is responsible for the next action?

Make that clear.

The AgentCentre AI Tools can help create and improve property management communication quickly.

They are particularly useful for:

• landlord updates

• renter responses

• maintenance communication

• difficult emails

• rent and lease discussions

• follow-up messages

• complaint responses

But the final step should always remain with the property manager.

AI creates the draft.

The PM checks the judgement behind it.

Before sending any AI-assisted communication, check:

Facts

• Names are correct

• Property details are correct

• Dates are correct

• Amounts are correct

• Quotes and contractor advice are accurate

• Nothing has been added that was not confirmed

Advice

• The recommendation reflects my actual view

• AI has not made a decision on my behalf

• Any risks have been explained accurately

• The advice is appropriate for this situation

Tone

• The message sounds like something I would send

• It is clear and professional

• It is not unnecessarily defensive or confrontational

• It is not overly long

• The tone suits the relationship and situation

Promises

• No unconfirmed outcome has been promised

• Any timeframe is realistic

• I can actually deliver what the message says I will do

• The next update date is achievable

Legal and Compliance

• Any legal reference has been independently checked

• The correct state or territory rules have been used

• Notice periods or deadlines are correct

• Formal wording or processes have been checked where required

Privacy

• No unnecessary personal information is included

• Internal comments have been removed

• Information about other clients or properties is not included

• Sensitive information is only included where genuinely necessary

Next Step

• The recipient knows what is happening

• Any required decision is clear

• Any required action is clear

• Responsibility for the next step is clear

• The next follow-up is clear where needed

Final Read

Before clicking send, ask:

If I had written this myself from scratch, would I be comfortable putting my name underneath it?

If the answer is yes, send it.

If not, edit it first.

That is probably the biggest risk with AI-generated communication.

It looks ready.

The spelling is correct.

The sentences are polished.

The tone appears professional.

So it is easy to click send without properly reading it.

Treat AI output as a first draft, even when it looks like a final one.

The PM still owns every message that leaves the agency.

AI can save the writing time.

It should never replace the final check.