20 Minute AI Appraisals Are Coming

[Non Sequitur]

Interesting blog, worth the time to read.

https://appraisalhouseusa.com/20-minute-AI-appraisals-are-coming

Link to the company mentioned, minimum $1,500 a month to set up and run. I'm sure they are the first of many, and word on the street is at least one AMC runs its staff side with similar software.

https://www.vervus.agency/

 

[shrubberyvaluation]

Interesting blog, worth the time to read.

https://appraisalhouseusa.com/20-minute-AI-appraisals-are-coming

Link to the company mentioned, minimum $1,500 a month to set up and run. I'm sure they are the first of many, and word on the street is at least one AMC runs its staff side with similar software.

https://www.vervus.agency/

Highly skeptical. Random numbers: Might get good enough to check boxes for FNMA.

50% Boost in efficiency.

50% increase in appraisal capacity.

 

[J Grant]

Interesting blog, worth the time to read.

https://appraisalhouseusa.com/20-minute-AI-appraisals-are-coming

Link to the company mentioned, minimum $1,500 a month to set up and run. I'm sure they are the first of many, and word on the street is at least one AMC runs its staff side with similar software.

https://www.vervus.agency/

AI can not do an appraisal, and idk if it ever could - it can assist, perhaps, in generating its own version of boilerplate and auto-fill sections of a report ( perhaps ). Of course, one would have to fact-check and logic-check every sentence.

It sounds like a money grab from this company - the ones who profit are tech guys who run companies that "trains" an AI for a hefty sum. .

There is a finite number of loans a month and thus a finite number of appraisals, so an appraiser can not make up in volume for a loss of appraisal orders or lower fees the tech might facilitate.

 

[ucbruin]

I don't get why folks think technology can't provide a credible appraisal....
Appraising is so simple that 70 year olds can still do them....

 

[Tom D]

AI can do an appraisal for only the middle of the bell curve in a general vagueness sense. But without the fine tuning of condition & location it not be accurate. Now the question is what % accurate variance would fannie accept. Fannie can do an accurate avm appraisal, cause the have every appraisal done in their system. But the accuracy is also based on when was the subject last seen. We know condition, or a neighborhood change, can make a big difference.

Every post here on what new is happening with appraising doesn't need AI to tell me what is happening. That sound you are hearing is more dirt on you coffin.
Dear AI,
What do the posts on appraisal forum tell me.
AI: Get the he^ll out of this business.

 

[Essexfenwick]

Only SRAs and MAIs in the end for the toughies

 

[J Grant]

AI can do an appraisal for only the middle of the bell curve in a general vagueness sense. But without the fine tuning of condition & location it not be accurate. Now the question is what % accurate variance would fannie accept. Fannie can do an accurate avm appraisal, cause the have every appraisal done in their system. But the accuracy is also based on when was the subject last seen. We know condition, or a neighborhood change, can make a big difference.

Every post here on what new is happening with appraising doesn't need AI to tell me what is happening. That sound you are hearing is more dirt on you coffin.
Dear AI,
What do the posts on appraisal forum tell me.
AI: Get the he^ll out of this business.

Seriously, AI can not do ANY appraisal, even in the middle of the bell curve. They talk about virtual AI "assistants," which are really glorified software functions that can save some time. My experience so far with AI is that it makes so many errors and wrong assumptions that the time to proofread and correct wipes out a lot of the time saved..

An AVM can spit out a value on any property in seconds, of course. Some of the AVM-generated values are very good, and others are not. The problem is that we can not tell the difference until other analysis is done.

 

[Vermonter]

Now the question is what % accurate variance would fannie accept.

Exactly.

Also have to figure in the cost reduction in other areas of the lending pipeline due to AI improvements. Originating, underwriting, risk analysis, etc. There's gonna be a tipping point and we won't know about it until it's over.

 

[Tom D]

Seriously, AI can not do ANY appraisal, even in the middle of the bell curve. T

Why can't AI use every free home value site on the internet to give you several values, and for free. Home owners don't care how, if they like the value being us or avm doing it. AI lending, 1 step shopping done in 5 minutes. Alexa can ask you the questions needed.

And i don't disagree with you at all J Grant. I post sometimes to rattle our coffins. Subliminally kinda, we are already dead.

 

[Non Sequitur]

Like it or not, a crap ton of money is flowing into anything AI right now and look at how VC and hedge fund money changed our industry in the past 5 years or so.

I don't think it's a stretch to train AI agents to crunch out boilerplate and filter MLS data for proximate and similar sales then sales to bracket features not bracketed or matched. What I see going forward is AI agents replacing the typists and virtual assistants the coach and team systems use now. And from what I hear that is well on its way and in some cases being used. But as J Grant mentioned there's a finite number of assignments, so the indie appraisers aren't in a position to pay $1,500+ a month or employ someone to develop a system. But the national firms and AMCs will and are jumping all in since they can lock up volume.