Jake HellerApr 20, 2026
Underwrite Deals in Your Own Model w/ Shortcut AI
Underwrite Deals in Your Own Model w/ Shortcut AI
Jake Heller
I test the Shortcut AI plugin to underwrite a multifamily real estate deal directly within my existing Excel model. While the tool is incredibly efficient at data entry and analysis, I highlight the critical need for human oversight and quality control.
Underwrite Deals in Your Own Model w/ Shortcut AI
Jake Heller
I test the Shortcut AI plugin to underwrite a multifamily real estate deal directly within my existing Excel model. While the tool is incredibly efficient at data entry and analysis, I highlight the critical need for human oversight and quality control.
Chapters
00:00
Introduction
Introduction to using Shortcut AI within an existing Excel underwriting model.
00:35
Setting Up the Analysis
03:36
AI Processing and Data Updates
05:46
Reviewing Results and Troubleshooting
11:20
Refining Assumptions and Quality Control
13:43
Conclusion and Future Outlook
Transcript
00:00
We're using shortcut AI to underwrite a new deal in our existing Excel model, our Pro forma.
00:09
Not another interface, not another tool.
00:12
This is our Excel model and we are going to use shortcut in it.
00:17
All right?
00:17
So at the top, right, we're going to use our shortcut Excel plugin, very easy to upload.
00:26
I'm already logged in here and you'll see it's a similar type of, interface that ChatGPT or Claude or any of these other large language models have.
00:35
what I am going to do is I'm going to take this deal, this om that Taylor Avakian had sent to us.
00:41
I'm going to attach it, okay?
00:43
And then I'm going to go to my AI, girlfriend, Claude and I already created a prompt, okay?
00:49
And I'm going to throw it in there.
00:50
It's pretty simple, but we want to perform a comprehensive underwriting analysis for the multi family property, etc.
00:56
Here's the critical instructions.
00:57
We want to use my existing Excel structure.
01:00
We want to populate the model with verified market data, replacing broker assumptions where incorrect.
01:05
And we do not want to create a new model.
01:06
We want to work within my existing framework.
01:08
So the use case here is that we're going to utilize the shortcut to underwrite using our own model.
01:16
And if you guys have been following along, you know that I really, really appreciate the tools that integrate pretty seamlessly with our existing workflows.
01:24
If you're an institutional shop, if you're any type of shop and you have a model just as this one, you see here, you don't want to abandon it, you want to use it, right?
01:32
You're comfortable with it, you know it inside and out.
01:35
It's what you've done.
01:35
We're stubborn in our ways, we don't like change.
01:37
And so what shortcut's so brilliant at and index, you could see I both have that, at the top here.
01:42
But they're so brilliant because they integrate seamlessly with our existing workflows.
01:46
So now watch us underwrite this deal.
01:48
We're going to hit send.
01:50
Let's watch shortcut go to work here.
01:51
And I haven't done this specific use case yet.
01:54
so we're going to give it a go and see what it's all about.
01:57
I have pretty high expectations, so I'm hopeful that it nails it.
02:01
Okay?
02:02
And what's great about shortcut too is you have a prompt library.
02:04
Okay?
02:05
So what I think these tools do is they level the playing field.
02:08
Okay?
02:09
They Level the playing field.
02:11
And I just want to check something here.
02:12
Where's my om?
02:14
So it's analyzing it.
02:15
Okay.
02:15
So it levels the playing field.
02:16
What I mean by that is, let's say, right?
02:18
So on a deal like this, this is pretty comprehensive stuff.
02:21
You know, underwriting a multifamily deal in la, it's no easy task, right?
02:25
But having shortcut or index or this Excel super agent at your disposal allows anybody to underwrite this type of deal.
02:35
Okay, There's a prompt library here.
02:37
So that exact prompt we could store in our library.
02:40
And then you could have your assistant, your View VA or whoever it might be, someone from your team who's maybe not the best underwriter.
02:46
They could take a stab at this.
02:47
They just input that prompt with the new Om and bang.
02:50
Shortcut underwrites the deal for you.
02:52
Now you need to be there for quality control.
02:55
Okay.
02:55
That'll never change.
02:57
Right.
02:57
Anytime you delegate tasks or things to another quality is, is, is something that needs to be thought of and considered.
03:04
but it levels the playing field.
03:06
The barriers to entry are less.
03:08
Okay, I could do this type of work for self storage, for retail, for multi, for industrial, for land, whatever it might be.
03:14
Okay.
03:15
I don't need to be an expert in underwriting.
03:17
Instead I just need to know the prompts and then I need to know how to direct shortcut and index and these AI tools to do what they do best.
03:28
So we'll report back shortly.
03:29
It looks like it's just studying the offering memorandum.
03:33
and I'll report back with some thoughts shortly.
03:36
So we're cranking.
03:36
So now it's going to start searching the web to verify market rents, operating expenses and comparables.
03:43
Watch this thing go to.
03:44
This is just nuts.
03:55
So now it is updating a lot of our information.
03:58
As you can see here.
03:59
Bang.
04:00
It's making lots and lots of changes.
04:04
We are getting some error messages, but I would think.
04:07
That it's going to fix all those for us momentarily.
04:10
It looks like it's updating the unit mix with 12 studios and 41 beds.
04:15
Let's take a look.
04:16
Bang.
04:16
There it goes.
04:17
Wonderful.
04:18
can figure this out.
04:18
It looks like it's trying to add gas, trash, and so it's reading this detailed utilities this detailed expenses sheet.
04:28
But now it looks like it figured out that it could only update utilities and it updated.
04:35
Looks like it's updating all of our growth assumptions here, which is pretty cool.
04:42
We are getting these error notifications quite a bit.
04:46
I'm not sure what that is.
04:48
Now it's updating our exit cap to 6.
04:49
Yeah, there's that error.
04:51
So Excel found a problem with one or more formula references in this worksheet.
04:55
Check the cell references, neighbor links to other workbooks and your formulas are all correct.
05:00
Yeah, I'm not sure what that is.
05:01
If you hit okay, it just moves.
05:05
So yeah, it's interesting.
05:06
So it's showing that our, our IRR has a value error.
05:11
So now it's going to look to see what the issue.
05:15
I see the issue.
05:16
The equity Multiple is showing 0.61 Whoops.
05:19
Which indicates negative returns at the asking price.
05:21
This confirms that we need to analyze the property.
05:25
So now it's running scenarios at different purchase prices.
05:29
It's created this analysis summary for us here, which I guess it's intending to Show U.S.
05:36
broker versus the verified assumptions in a comparison table.
05:40
Interesting.
05:41
Here we go.
05:42
So it's building all that out for us.
05:46
So we got this pretty beefy analysis it's doing on a separate tab here.
05:51
And so it's showing us.
05:53
Broker versus verified.
05:57
Look at all these strengths and weaknesses it's identified.
06:02
Wow, this is awesome.
06:03
Strong Silver Lake location, major Systems renovated in 2020.
06:07
Multiple units with huge upside to market.
06:10
And then we have lots of weaknesses and risks.
06:12
Rent control, 1916 building insurance costs likely understated and rising rapidly.
06:18
No parking limits.
06:20
Tenant pool renovation cost estimated at 15 to 20k per unit.
06:24
Value add strategy.
06:26
That's awesome.
06:27
So it's conclusion.
06:29
So its conclusion is do not proceed at asking price of 3.1 million.
06:35
It recommends submitting an offer at 2.75 million with thorough due diligence contingencies.
06:42
Wow, look at this formatting and everything I put together here looks pretty good.
06:45
It's got some sales comparables here.
06:48
Market rent comparables.
06:50
Now it's testing different scenarios at different purchase prices.
06:54
But we're still getting this value error here and I want to see if it picks up on that again.
07:00
Now it's trying 2.2 million You know, it's interesting because they may have changed our formulas up here.
07:09
They may have changed our formulas.
07:12
So that's not great.
07:15
Yeah, I think did something to formulas.
07:19
We should get a positive IRR and equity multiple.
07:23
Especially I mean, if our average cash on cash is 8% what purchase price?
07:27
That's 3.
07:28
Yeah, I think it did something to our.
07:30
Our returns.
07:31
I mean, it.
07:32
Look, it did a great job with all this.
07:34
Let's scooch that over and you can see, like, now it's giving us an executive summary.
07:40
I mean, the way this all gets formatted and done is like, insane.
07:44
Look how ridiculous this is.
07:46
Look at this property represents a value trap, not a value add opportunity.
07:51
This is such a detailed analysis on top of our actual model.
07:55
So now it looks like it's building a.
07:58
Building something.
08:00
Where is that?
08:01
It's building a detailed rent roll analysis.
08:07
There it is.
08:08
So it just built this detailed rent analysis for us, which is pretty cool.
08:14
And it looks like it's making this analysis summary, the main sheet.
08:18
So it's showing current and then market rents and then monthly upside priority of turning units.
08:26
Pretty remarkable, actually.
08:27
so it's definitely not perfect with our actual underwriting model with that use case.
08:35
I mean, that was a pretty complicated thing to do, right?
08:39
Take an om.
08:40
go verify all the assumptions, figure out how to use our model, input all of those assumptions into the right locations.
08:47
I mean, look, this model is like really, really robust and comprehensive.
08:51
So the fact that it got this far, I think is impressive.
08:54
Now I just want to see why.
08:56
I think there's an error.
09:02
And that error keeps popping up too.
09:04
That's kind of annoying.
09:04
I think there's an error with the IRR and equity multiple return calculations.
09:13
Can you please try and figure this out?
09:18
I.
09:19
I think we should have a positive IRR and equity multiple.
09:27
And it appears the formula or calculation is now incorrect.
09:36
Let's see what happens with that.
09:38
And here's our fully kind of comprehensive summary, Final recommendations.
09:43
I mean, it says, do not proceed at any price.
09:45
This property represents a value trap.
09:48
Maybe I am wrong.
09:50
And let's just see what it kind of does here.
09:54
I hate that error thing that keeps popping up.
09:56
That's really annoying.
09:57
So now it's going to investigate this calculation here.
10:01
Let's see what we can find out.
10:07
So it looks like it may have found the error here.
10:09
my God.
10:10
Look, it fixed it.
10:11
That's unreal.
10:13
Okay, so that's what I thought.
10:15
So now all of our returns makes sense.
10:17
Okay.
10:18
All right.
10:19
So yeah, we need.
10:20
There needs to be a little babysitting in this scenario and that.
10:23
That's fine.
10:23
Okay.
10:24
But, now it needs to update update everything and look again.
10:28
That's okay.
10:28
I Wonder why.
10:29
I wonder if it changed our formulas for that, or what exactly happened.
10:35
But like now, now we got something here.
10:38
So I'm glad it was able to figure that out.
10:40
average cash on cash, five and a half percent.
10:42
I mean that, that makes sense.
10:44
spread between your untrended yield on cost and exit cap.
10:49
Less than 100bps.
10:51
Not great.
10:52
18 gross levered IRR feeling, is maybe that's low.
11:00
and let's just see what else our rents.
11:04
I mean it, look, it says it verified all of that.
11:07
Our expenses were at 36%.
11:09
Yeah.
11:10
So that, that's probably the issue or the expenses.
11:12
It's, this is honestly so amazing like how it does this.
11:16
I mean having a co pilot like this is just so insane.
11:20
I mean these expenses need to be a lot higher.
11:23
So I, I, what I would say is that it kind of missed on that research.
11:29
This is a 1916 building.
11:31
We need to be closer to like 45.
11:33
And then I think like repairs and maintenance for 1916 building should not be 500 per unit per year.
11:41
Like I mean I think a thousand's like a lot closer to that.
11:46
And you'll see that the returns kind of reflect that we're still only at 39.
11:51
Let's just see payroll.
11:53
Like that's fine.
11:54
It's a 16 unit building.
11:55
General admin.
11:56
Fine utilities.
11:57
That seems really low.
11:58
So I don't know what's going on with that insurance.
12:00
I mean probably even like 1500.
12:03
honestly, 1500 per unit minimum LA, like 1916 building.
12:08
Come on, seriously.
12:10
So now we're starting to get into.
12:12
and that's why like this, this, you know, is it going to replace the analyst?
12:16
No, like you know, the analyst is there for exactly what I'm doing right now.
12:21
Okay.
12:21
And that's quality control.
12:23
but actually building this out, taking a first stab at it, we didn't have to sit here and watch this, you know, we could have been doing other stuff.
12:29
So I think that's what's so brilliant about this.
12:31
Right?
12:31
Like maybe the 10 to 15 minutes I spend doing quality control, you know, that allows me to underwrite a lot more deals versus having to do data entry and all this garbage and initial, initial research.
12:42
Right.
12:42
A lot of that could be done with the AI, which I think is so cool.
12:45
But it definitely missed on some assumptions here.
12:48
that's fine.
12:48
I mean, look, even 41, I'm, this should be a 5% management fee on a small building like that.
12:55
Utilities.
12:55
I Just can't see how it's only 650, but maybe we're getting real numbers.
13:01
Make ready contract services, all that's fine.
13:04
Property taxes appear to be right.
13:06
Capital reserves, come on.
13:07
350 a unit minimum.
13:10
Like, let's be honest.
13:11
And then is our exit cap really a six on a 1916 building here?
13:17
No, probably make that six and a quarter.
13:19
So now our returns are kind of more in line.
13:22
So that's at the, the listing price, right?
13:24
So like, where does this deal get done?
13:27
Maybe there, maybe not.
13:28
But so anyways, it's updated everything here.
13:31
It's updated its full analysis summary.
13:34
Proceed.
13:34
with due diligence.
13:35
Property shows strong returns.
13:37
Boom.
13:37
Boom.
13:38
Wow.
13:38
Look at like, this level of detail is so insane.
13:41
So human judgment, quality control.
13:43
More of the story here, guys.
13:45
We're not quite there yet.
13:46
With fully, automated, underwriting.
13:50
We're close.
13:51
Not quite there yet.
13:52
I'm going to do probably the same exact demo using Index and I will compare the results and see which one is doing a better job.
14:01
so excited to do that.
14:03
This was a fun one.
14:05
I certainly learned a lot.
14:06
And I learned that this Excel agent is absolutely brilliant and can be utilized in such efficient and strategic ways.
14:16
You cannot, rely on it to underwrite deals utilizing your model without oversight, direction and quality control.
14:27
That's sort of the moral of the story here.
14:29
But the way that it's created these formulas and does its homework and research, and then builds a separate tab to do comparison and a full analysis, summary and formats it and just, it's so brilliant and just wicked, wicked intelligent.
14:48
there is no analyst who could perform at the abilities of these Excel agents.
14:55
There is none whatsoever.
14:58
So super, super excited to be, you know, working with Shortcut to be working with NDEX as it relates to real estate, to making them better.
15:08
giving real time feedback, getting you guys all access to this stuff, it's just so awesome.
15:14
So I'll be doing a demo, same sort of demo with NDEX just shortly.
15:20
and I, hope you guys enjoyed this.
15:22
Let me know if you have questions.
15:23
I'm going to keep doing different use cases with shortcut index just to see like, where it performs better.
15:30
but using your own model template, which in this case mine is crazy robust.
15:37
If mine was a lot simpler, it'd probably do a lot better job, I think shortcut.
15:42
But because this is so robust and complicated, which again, I mean, like, look at, Let me just show you guys, like look how insane this model is, I mean, like, all these different cash flows and stuff.
15:54
It is so complicated.
15:56
So, I mean.
15:57
And it goes forever to look at all this.
15:59
This is insane.
16:00
So the fact that shortcut was able to work within that model, as it just did, I think is so impressive.
16:07
So we're close.
16:08
We, are close, guys.
16:10
We're very close.
16:11
But we're still needed for quality control and judgment.
16:16
I don't think that'll ever go away.
16:18
And so the best analysts, in my opinion, are, going to be the ones who could effectively use these tools and multiply how many deals they could underwrite.
16:26
So we're going to figure out how promise.