How to Build an AI Underwriting Workflow for CRE

I run the AI for CRE Collective (540+ members testing AI tools on real deals), and the number one complaint I hear about AI underwriting is consistency. The outputs aren’t standardized. So I built a custom underwriting skill inside Manus that uses my exact Excel model. Every deal gets the same template, same tabs, same formulas — with deal-specific numbers plugged in.

Most CRE acquisition teams still spend hours manually entering OM data, rebuilding underwriting models, and formatting Excel files for every deal. AI underwriting workflows reduce repetitive manual work and improve underwriting consistency across acquisition pipelines.

Why Most AI Underwriting Workflows Fail

This lack of standardization is one of the biggest challenges with AI underwriting for CRE workflows today. Ask Manus (or any AI) to “underwrite this deal,” and you’ll get an Excel file. It’ll have some financial assumptions. And it’ll look completely different from the last one it produced. Run it 10 times, and you’ll get 10 different structures. Some have sensitivity tables and calculate IRR. Others skip it. You can’t hand these to an LP or plug them into your existing workflow. The fix is skills — explicit instructions that tell the AI exactly how to structure the output every time.

What You Need to Build an AI Underwriting Workflow

How AI Underwriting for CRE Improves Deal Standardization

Before building the skill, you need:

1. Your underwriting model in Excel.
   The one you actually use. Not a template you found online. Your actual model with tabs, formulas, and sensitivity tables.

2. A Manus account.
   The skills and the agents feature both need to be active.

3. About 20–30 minutes.
   The upfront investment pays off on every deal after this.

Step 1 — Upload Your CRE Underwriting Model

Go to Manus settings and find the skills section. Hit “Build with Manus” and tell it what you want:

“Help me create a skill. Anytime I ask you to underwrite a deal. to underwrite it using my exact model attached here. It should use this exact template with updated numbers from each deal. I will usually send you an OM and ask you to underwrite. You will go ahead and underwrite the deal and return this exact model with the updated numbers for that specific deal.”Then upload your Excel file.

I added one more line that made a difference:
“Please ask me questions if you aren’t clear.”That last part is important. It forces Manus to clarify before it makes assumptions.

Step 2 — Train the AI Underwriting Workflow

Manus will go into your Excel model and ask specific questions. When I did this, it asked:

• Should sensitivity tables be recalculated, or should the deal’s inputs be hard-coded values? (Recalculate them.)
• Should it implement an actual solver to find the purchase price? (Yes.)
• Should it auto-generate analyst notes based on what it extracts from the OM? Yes.

Step 3 — Build the AI Underwriting Skill

Manus will initialize the skill and build all the resources. It maps every tab in your Excel file, creates reference documents for the model specs, and writes the full skill instructions. This took about 10–15 minutes in my test.

You’ll see it working through subtasks — mapping the template, cleaning up example files, copying the model, and creating extraction docs. When it’s done, the skill shows up in your skills library. Mine is called “Multifamily Underwriter.”

Step 4 — Test the AI Underwriting Workflow With a Real Deal

Here’s where it gets good. I attached an OM for a property at 163 8th Place, texted Manus from Telegram, and said:

“Please underwrite this deal using our new multifamily underwriter skill.”Manus extracted the deal data from the OM, researched market comps, validated assumptions, and built the full pro forma using my exact template. The result: 5.4% IRR at the asking price. Recommendation: pass. The returns were below typical institutional targets for a value-add deal. And the model looked exactly like mine. Same tabs.  structure and sensitivity tables. With the deal’s actual numbers populated throughout.

What Worked Well

The standardization is the real win. Every deal that goes through this skill comes back in the same format. You can compare them, stack them, hand them to your team or investors, and they know exactly where to look.

The agent also self-corrected when it hit a snag. At one point, it couldn’t find the skill files. Without me saying anything, the agent went back, located the files, and sent them to the subtask. Then it finished the underwriting.

AI Underwriting vs Manual CRE Deal Analysis

Process Stage	Manual CRE Underwriting Workflow	AI Underwriting Skills in Manus
OM Data Extraction	Manual data entry from PDF	Automated OM data extraction
Excel Model Setup	New file for every deal	Standardized Excel template
Assumptions Entry	Manually input each time	Auto-filled from OM
Sensitivity Tables	Rebuilt for each deal	Automatically recalculated
IRR & Return Metrics	Analyst-dependent	Model-driven outputs
Deal Comparison	Difficult across files	Easy side-by-side analysis
Formatting Consistency	Varies by analyst	Same tabs and structure
Time Per Deal	1–2 hours	20–30 minutes
Lender/LP Review	Needs formatting edits	Ready-to-share output
Workflow Scalability	Limited by manpower	Scales across deals

Comparison between manual CRE underwriting workflows and AI underwriting skills in Manus for standardized deal analysis.

How AI Underwriting Automates CRE Workflows

AI underwriting workflows automate many repetitive commercial real estate underwriting tasks that traditionally require manual analyst review. Instead of rebuilding spreadsheets and reviewing offering memorandums line by line, AI underwriting systems can process financial data, extract documents, and standardize underwriting outputs automatically.

Modern AI underwriting workflows support:

• underwriting process automation,

• document processing,

• document extraction,

• workflow automation,

• financial statement analysis,

• balance sheet review,

• ratio analysis,

• underwriting decisions,

• risk assessment workflows.

This improves consistency across acquisition pipelines while reducing repetitive manual underwriting work for CRE analysts.

How AI Underwriting Uses Machine Learning and AI Models

Modern AI underwriting platforms combine artificial intelligence, machine learning, large language models, and AI-powered automation to improve underwriting workflows and deal analysis.

These AI underwriting models can:

• analyze data points,

• identify risk signals,

• review financial data,

• automate document processing,

• support underwriting decisions,

• improve predictive analytics,

• standardize underwriting outputs.

Instead of manually reviewing every acquisition package, CRE teams can use AI systems to automate underwriting tasks while improving workflow scalability and operational efficiency.

Why CRE Teams Are Adopting Automated Underwriting Systems

Commercial real estate firms are increasingly adopting automated underwriting systems to improve acquisition speed, workflow consistency, and portfolio performance analysis.

AI-powered underwriting systems help teams:

• automate underwriting workflows,

• process financial statements faster,

• improve deal comparison,

• reduce analyst workload,

• standardize underwriting outputs,

• improve document automation,

• scale acquisitions operations.

While human underwriters still review assumptions and final outputs, AI underwriting systems dramatically reduce repetitive manual work across CRE acquisition pipelines.

Where It Falls Short (For Now)

The underwriting isn’t to be 100% accurate out of the box. You still need to verify assumptions, check rent comps, and validate the numbers. This is a first draft, not a final product. And building the skill takes real time upfront. But that’s the whole point — invest 30 minutes now, save hours on every deal after.

FAQs regarding AI Underwriting for CRE in Manus

Why does AI underwriting for CRE produce different Excel outputs each time?

AI underwriting for CRE often gives inconsistent Excel models because it lacks a standard structure.

• Generic AI tools build a new format for every deal

• Tabs, assumptions, and formulas may change

• Sensitivity tables may appear or disappear

• IRR and DSCR models are not always included

This makes deal comparison across multiple CRE investments difficult.
Learn more about AI in financial modeling: https://www.mckinsey.com/capabilities/quantumblack/our-insights

How can Manus standardize AI underwriting for CRE workflows?

Manus can standardize AI underwriting for CRE using custom skills.

• Upload your Excel underwriting model

• Train Manus to follow your structure

• Map deal inputs to your formulas

• Use the same tabs for every deal

This ensures all CRE deals return in the same Excel format.
Explore underwriting automation basics: https://www.investopedia.com/terms/u/underwriting.asp

What is a custom underwriting skill in Manus?

A custom underwriting skill is an AI SOP for CRE deal analysis.

• It uses your actual Excel model

• Keeps the same sensitivity tables

• Applies your IRR and return formulas

• Updates numbers for each OM

This helps automate CRE underwriting without changing templates.

Can AI underwriting for CRE read offering memorandums automatically?

Yes. AI underwriting for CRE can extract deal data from OMs.

• Reads the rent roll and expenses

• Pulls unit mix and income

• Maps assumptions to your model

• Updates financial projections

This reduces manual data entry for CRE analysts.

How does AI underwriting for CRE improve deal comparison?

Standardized AI underwriting for CRE improves deal comparison.

• Uses one Excel template

• Keeps the same assumptions

• Aligns IRR and cap rate inputs

• Maintains formula consistency

This allows side-by-side CRE deal evaluation.

Do CRE professionals still need to verify AI underwriting outputs?

Yes. AI underwriting for CRE creates a first draft only.

• Rent comps may need validation

• Expense ratios require checks

• Market assumptions may vary

• Renovation costs may differ

Manual review ensures underwriting accuracy.
Review financial due diligence: https://www.pwc.com/gx/en/services/deals.html

How long does it take to set up AI underwriting for CRE in Manus?

Setting up AI underwriting for CRE in Manus takes about 20–30 minutes.

• Upload your Excel model

• Answer skill questions

• Map underwriting tabs

• Test with a real OM

This upfront setup saves hours on each deal later.
Explore workflow automation: https://zapier.com/blog/workflow-automation/

Can AI underwriting for CRE support multiple unit types?

Yes. Manus can support multiple unit types in CRE deals.

• Handles studio to multi-bed units

• Applies rent assumptions

• Updates renovation premiums

• Adjusts revenue projections

This supports multifamily underwriting workflows.

The Bigger Picture

Skills are basically SOPs for AI. However, your analyst would screen a deal, map it out step by step. Give those instructions to the AI. It builds the skill. Then you trigger it by text from your phone. I’ve been doing this with Claude for months. Manus adding the same concept, plus the Telegram agent layer, is a huge step forward.

If you want to see the full demo video and the actual underwriting output, I shared everything within the AI for CRE Collective. We’ve got 540+ CRE professionals testing workflows like this every week. For weekly CRE AI breakdowns and tool updates, you can also join our newsletter, followed by 3,000+ industry professionals.