Free playbook for commercial real estate investors

The Complete Claude Playbook: Commercial Underwriting

A commercial asset is a stack of long, dense contracts — a 60-page lease, three amendments, two estoppels, a rent roll that does not foot — and a human has to read every one before the math means anything. This playbook hands Claude the heavy reading and the first-pass underwrite, so your hours go to the judgment calls that decide the deal.

One PDF, end to end: the four-stage engine, the per-asset signal library, the page-cited abstraction method, the scoring model, and the honest ceilings. Built for Claude across NNN, retail, office, and industrial. The first read of a single package can swallow an analyst's week — this is how you get it done by morning.

Download the pack

Built around your acquisitions.

Get the full pack once, then use the walkthrough below to see exactly what is inside and how to run it.

Get The Complete Claude Playbook: Commercial Underwriting

Enter your details, then download the full playbook PDF. No cost.

Want this running on your deal flow?

Sourcing, owner resolution, underwriting and documents. Choose your modules, configured around your firm.

Book a demo

Who this is for

If your pipeline runs on deal packages, the constraint isn't capital or access to listings. It's that every asset is a stack of long contracts a human has to read before the math means anything. This removes the reading as the constraint.

Acquisitions principals

An OM hits your inbox at 4pm with indications due Friday. You triage on gut and the broker summary because there's no time to triage on the leases. This gives you the read by morning.

Net-lease and retail buyers

True NNN or quietly NN, co-tenancy traps, kick-out rights, percentage rent — the value and the risk are written down. The playbook abstracts every term with a page cite so you underwrite the lease, not the summary.

Funds watching a portfolio

The deal is the lease stack and the debt clock. The Monitor stage watches your owned-asset maturity wall, so a balloon into a higher rate surfaces before it becomes the seller's problem.

The four-stage engine

Ingest, Abstract, Underwrite, Monitor. Each stage is a production prompt with input slots and a fixed output contract, so the package flows through without losing terms. Every extracted term carries a page cite back to the source PDF, and a human gate sits before any number enters a model.

Ingest the leases, amendments and estoppels

Take the raw deal package — OM, every lease and amendment, estoppels, rent roll, debt summary, tenant list — and normalize the mess. OCR the scanned leases, reconcile the rent roll into structured rows, and flag what's missing. Nothing silently dropped.

Abstract every term with a page cite

Read each lease into structured terms: commencement and expiration, base rent and rent steps, renewal options, recovery type (true NNN vs NN vs modified gross), TI, free rent, co-tenancy and exclusive-use clauses, guarantor. Every term carries a page cite. No cite, no term.

Underwrite cap rate, WALT and tenant concentration

Roll the abstracted leases into a first-pass model: in-place NOI, cap rate, WALT, recovery leakage, the rollover schedule, and how much of the NOI rides on the top tenant. The cap-rate and WALT view a principal needs to say pursue or pass.

Monitor the maturity wall on what you own

For owned assets and a tracked pipeline, watch the debt maturity wall over time — balloon proximity, watchlist and special-servicer status, occupancy decline, tenant credit downgrades. Surface the asset before the refinance becomes the seller's problem.

Tune the per-asset signal library

Green and red underwriting flags, tuned per asset type: NNN credit and true-net structure, retail co-tenancy and kick-out clauses, office rollover cliffs, industrial functional fit. Additive and explainable, so you can always read why a deal ranked where it did.

Keep the honest ceiling honest

AI does the heavy reading and the first-pass underwrite, not the judgment. Abstraction is first-pass, not straight-through; private-tenant credit is opaque; cap rate and WALT are only as honest as the rent roll you feed in. The legal read and the IC call stay human.

How to run it

Start in conversation mode today: a Claude project, the four-stage prompts, and one real deal package. Agent mode is the production configuration that watches your portfolio's maturity wall every night, and the playbook doubles as its specification if you have an engineer.

Claude

Project, conversation mode

No code

Claude Code

Agent mode

Dev env

Your underwriting model

The engine fills it

Any

Your first run

Drop one real lease, its amendments, and the estoppel into a Claude project and open the Abstract stage:

You are abstracting a commercial lease for an acquisition underwrite.

LEASE: [attach the lease + all amendments and the estoppel]
ASSET TYPE: [NNN / retail / office / industrial]

Extract every term into a structured abstract: commencement and
expiration, base rent and rent steps, renewal options, recovery
type (true NNN vs NN vs modified gross), TI, free rent, co-tenancy
and exclusive-use, guarantor.

Cite the page for every term. If a term isn't in the document, say
"not found" — never infer it.

The playbook walks all four stages, the per-asset signal library, and the scoring model, plus the human checkpoints that keep an abstract an operational summary, not a legal opinion.

Built around your acquisitions.

Choose your modules. Configure them around your firm.

Book a demo Join the community