A colleague mentioned last week that she’d cut her email drafting time in half using ChatGPT. My first reaction was skepticism—wouldn’t AI-written emails sound generic and impersonal?

After testing it myself for a month, I’m now somewhere between “this is useful” and “this requires careful management.”

What’s Actually Working

Initial buyer enquiry responses. When a new enquiry comes in, AI can draft a professional response that includes property details, inspection times, and next steps. I feed it the listing description and say “respond to this buyer enquiry,” and it produces something usable in seconds.

The key is customisation. The AI draft is a starting point. I add personal touches, specific details about what caught their interest, and anything relevant from previous conversations.

Post-inspection follow-ups. “Write a follow-up email to a buyer who inspected [address] yesterday. They seemed interested in the kitchen renovation and asked about the strata fees.” The AI handles structure and tone while I add the substance.

Vendor weekly updates. The structure of these emails is often similar—market activity, inspection numbers, buyer feedback. AI can create the framework quickly, and I fill in the specific details.

Declining to proceed emails. These are awkward to write. AI can help with professional, compassionate language that gets the message across without burning bridges.

Where It Falls Short

Anything requiring local knowledge. AI doesn’t know that the train line extension will affect this suburb in 18 months, or that the neighbour’s DA was rejected last year. Local context matters, and AI can’t provide it.

Emotional situations. Deceased estate vendors, divorcing couples, stressed first-home buyers—these situations require genuine human sensitivity. AI-generated sympathy reads as hollow.

Negotiation communications. Too much is conveyed through careful word choice and tone. I wouldn’t trust AI to draft an offer response or counter-offer communication.

Anything where accuracy is critical. AI occasionally makes things up. It might insert incorrect property features or imaginary nearby amenities. Everything needs to be fact-checked before sending.

The Workflow That Works

Here’s my current process:

1. Open ChatGPT (or Claude, which I sometimes prefer for longer content)
2. Paste in relevant context: listing details, previous emails, specific situation notes
3. Give a clear instruction: “Draft a professional email to [recipient type] about [topic]. Tone should be [friendly/formal/warm].”
4. Review the output, checking for accuracy and tone
5. Add personal touches and local knowledge
6. Remove anything that sounds generic or corporate
7. Send

Total time savings: about 50% on routine correspondence. More time saved on emails I was procrastinating on because I didn’t know how to start.

The Risks

Over-reliance. If every email sounds the same, you lose personal brand. Buyers and vendors should feel like they’re communicating with you, not your AI assistant.

Errors slipping through. The more you trust AI output, the less carefully you check it. One incorrect detail in a buyer email can undermine credibility.

Privacy concerns. Be thoughtful about what information you paste into AI tools. Personal client details, financial information, and sensitive situations probably shouldn’t go through third-party AI services.

The Bottom Line

AI email drafting is a time-saver for routine communications. It’s not a replacement for genuine human connection in high-stakes conversations.

Use it for structure and efficiency. Keep the personal touch, local expertise, and emotional intelligence firmly in human hands.