CRM & leads

AI in your CRM: what helps and what's hype.

Every CRM now says it has AI. Some of it saves hours a week. Some of it is a button that writes a generic email you'll delete. Here's how to tell the difference, and the guardrails any AI should have before it speaks to your clients.

The test: does it remove a task or add one?

Useful AI removes something you were doing by hand. Answering a lead at 10 PM, summarizing a long email thread, drafting a reply you'd have written anyway. Hype AI adds a step: a "generate" button whose output you have to read, fix and paste, taking longer than writing it yourself.

When a vendor demos an AI feature, ask: what did I do before this, and what do I do now? If the honest answer is "the same thing plus checking the AI," skip it.

After-hours replies

This is where AI earns its keep in real estate. Leads arrive at night and on weekends: open-house sign-ins on a Sunday, ad leads after dinner, buyers scrolling listings at midnight. An assistant that replies within a minute, answers simple questions from your information, and offers times on your calendar turns those into booked calls instead of cold leads by morning.

The difference between good and bad here is the handoff. A good assistant knows when it's out of its depth and passes the conversation to you with a note. A bad one keeps talking.

Drafted replies and summaries

For conversations you handle yourself, AI that drafts a reply in your voice from the thread can save real time, as long as you read it before sending. Summaries help most on long, multi-party threads like transaction email, where reading the gist in two sentences beats scrolling through twenty messages.

Look for drafts that learn from your edits. If you keep changing "Hi there" to "Hey," the drafts should eventually start with "Hey."

Voice: making AI sound like you

Clients can tell when a message wasn't written by you. The fix isn't avoiding AI; it's giving it your voice. Good tools learn from samples of your writing and from the edits you make to their drafts. Some let you record a short talk about how you work and pull your real phrases from it.

Give it material. Paste in a few texts you've sent to clients, a past newsletter, an email you're proud of. Tell it what you never say. If you hate "Just checking in!", put that on a banned list. Then judge the drafts for a week and keep editing. If they still sound like a stranger after that, the tool isn't learning.

And keep some messages entirely your own. A note after a closing, a condolence, a congratulations: those should come from you, typed by you.

Features that sound good but rarely help

  • Lead scores with no explanation. If you can't see why a lead is a 92, you won't trust it.
  • Fully automated nurture written by AI. Long-term messages should be written once, carefully, by you.
  • AI "insights" dashboards that tell you things you already know.
  • Chatbots on your website that can't book a call or hand off to you.

The guardrails to insist on

Before any AI talks to a client on your behalf, confirm in writing that:

  1. It starts in a review mode where you approve its replies before it sends on its own.
  2. It never quotes prices, rates, commissions or legal terms.
  3. It never gives legal, tax or lending advice.
  4. It hands off to you when unsure, and tells you why.
  5. It doesn't make up facts about a property; it only uses information you've provided.
  6. You can see every message it sent, with a timestamp.
  7. It respects opt-outs instantly.

Fair Housing matters here too. An assistant should describe properties, not people, and never steer anyone toward or away from a neighborhood. You remain responsible for what goes out under your name.

Privacy and data questions

AI features send your conversations to a model to process them. Before you switch them on, ask the vendor a few direct questions:

  • Which AI provider processes my data, and is it used to train their models?
  • Are sensitive numbers, like account or Social Security numbers, masked before anything is sent?
  • How long are conversations stored, and can I delete them?
  • Can I turn AI features off for specific contacts or channels?

Clear answers are a good sign. Vague ones are a reason to keep the AI features switched off until you get clarity. Your clients trust you with personal information, and that trust extends to the tools you choose.

Switching on AI without surprises

Switch on AI features one at a time, starting with the ones that only draft and never send. Use drafted replies in your inbox for a week and notice how often you change them. When the drafts are mostly right, consider after-hours replies, starting in review mode so every message waits for your approval. Read the conversations each morning. When you've gone a week or two without needing to change anything, let it reply on its own for simple questions while still handing off anything unclear.

Tell your team what's switched on. Nothing confuses a client faster than an AI reply followed by a colleague's message saying something different. And keep reviewing. A monthly read-through of what the assistant said catches small problems before they become a client complaint.

Measuring whether it's working

Give any AI feature a simple scorecard. For after-hours replies, track how many late leads got an answer within a few minutes and how many became booked calls. For drafted replies, notice whether you're sending more messages in less time. For summaries, ask whether you're catching problems earlier in your transactions. If after a month you can't point to a concrete change, the feature isn't pulling its weight, and it's fine to switch it off. AI should earn its place in your workflow the same way an assistant would.

How KONAH uses AI

KONAH CRM 2.0 Pro ($149.99 a month) includes KONAH Assistant, which drafts replies in your voice; you read and send them. Its AI receptionist is coming soon. It never quotes prices or rates, never gives legal or tax advice, never takes payment details, and hands the conversation to you with a morning task when it isn't sure. The standard KONAH CRM 2.0 ($69.99 a month) includes AI-drafted replies in the inbox; nothing sends without you.

Put follow-up on rails. KONAH CRM 2.0 texts and emails every form lead in seconds and asks for a review when a deal closes.

Questions, answered.

Will AI replace my assistant?

Not the judgment. It can take routine after-hours replies and drafting off their plate so they spend time on the work that needs a person.

Can AI replies hurt my reputation?

They can if they guess or overpromise. Insist on review mode, a clear handoff, and an assistant that only uses facts you've given it.

Is AI-written content a Fair Housing risk?

It can be if it describes people rather than properties. Read what it drafts, and use tools that flag risky wording.

See KONAH CRM