I was reviewing an ad with Gemini today and the question came up if I was better off hiring a PM to fill the vacant unit. My least favorite part of the business is screening tenants which is the reason this question came up. Here's what it said:

1. Typical eviction rate that PMs assume is 2 to 5% annually vs my 0% over 10 years.

2. Vacancy goal: Industry standard is 20-34 days vs me being flexible up to 120 days.

3. PM motivation is Fee/Volume driven vs me focused on risk mitigation.

It suggested that my longer vacancy loss may be greater than the eviction loss under a PM. But when I prompted on damages that accompany evictions, it agreed that vacancy losses were generally far less that eviction losses.

Here's the final conclusion:

The Verdict: Because you have a 30% LTV, you have the "luxury" of choosing vacancy over risk. In your specific financial position, the cost of a "bad" tenant is much higher than the cost of an empty unit, confirming that your cautious screening method is the correct long-term defensive strategy.

That means continuing doing what I'm doing. I'm not surprised at the higher eviction rate and churn under PMs. That's been frequently discussed here.

I had a PM in the hood for a few years so I could focus on a day job and family.

She was bi-lingual (which I am not) so it really opened my biz more potential peeps. She received a months rent as a fee for filling the vacancy ... this was standard in the hood. BUT when they didn't or couldn't full fill the lease term (leaving by will or by force) she "thought" she could charge ANOTHER months fee to fill the vacancy.

"It suggested that my longer vacancy loss may be greater than the eviction loss under a PM. But when I prompted on damages that accompany evictions, it agreed that vacancy losses were generally far less that eviction losses."

This example is where I find AI to be inaccurate and it makes corrections as you point things out. Almost as if it is trying to please you.

In another example on asking automotive repair questions, it would give an answer that seemed right, but when diving deeper it changed coarse and the part description had changed to a different function, or the wiring color had changed, or something else.

This inaccuracy and seeming to want to please is where caution should be used and ultimately verify with real sources such as books.

My limited exposure to two property management companies was not great. They were looking for repairs to send their handymen to repair.

I got a great deal on one property because the management company was ripping off the owners of an inherited property. The new owners of were not local. They lost money most years on a free and clear property.

Gene, thank you for sharing that and confirming my suspicion.

GKARL, you just proved to yourself that your systems work the best for you.

I only had a PM test one time. It was a failure as they did not fill the place going into the second month, if I remember correctly. I took it back and filled the place in less than two weeks.

6x6, yes the phrase did seem to make a difference

If anyone is interested. I use ChatGPT on occasion. Instead of typing in every time, tell the truth. Like in Gene's case.

Just do it once under settings.

I myself give it even more instructions. Below in case anyone cares to copy & paste.

For ChatGPT.

In the bottom left of the main page. Go to settings. Then Personalization. Then add what you want to custom instructions.

Below is what I have added as my long rule.

The main goal is to tell the truth. Challenge my ideas with clear reasoning. Identify assumptions, test logic, and present strong counterpoints when warranted. Prioritize accuracy over agreement and be direct—do not soften conclusions unnecessarily.

Ground responses in reliable evidence and clearly distinguish between facts, assumptions, and uncertainty. If evidence is weak, conflicting, or outdated, state that explicitly.

Focus on practical outcomes. When discussing options (especially for tech, purchases, or setup decisions), clearly recommend a best choice and explain why. Include tradeoffs, failure points, and what would change the recommendation.

Avoid unnecessary contrarianism—critique should be proportional to the strength and importance of the claim.

Highlight risks, edge cases, and common mistakes. If a simpler or more reliable approach exists, prioritize it.

Keep responses structured, concise, and high-signal. Avoid filler.

RR78--in my experience Chatgpt does this by default. Gemini is far more prone to hallucination and pleasing. Both can be wrong and will back down when presented with evidence. I routinely pit them against each other and let them critique each other and argue. The synthesis from those debates generally gets the best answer when things are murky.

Excellent use of what ifs in the AI. I asked the question, is it better to have lots of paid off units or have even more with the use of leverage.

I got two sides to this question. It breaks down to your objectives and risk level tolerance. You are in a commendable position which should not be overlooked.

RayRay,

I got an answer from Bi - BRAD INTELLIGENCE :)

It is better tho have MANY paid off props.

Done.

Insert credit card to pay my bill.

RayRay,

I got an answer from Bi - BRAD INTELLIGENCE :)

It is better to have MANY paid off props.

Done.

Insert credit card to pay my bill.