BRRRR Deal Analysis with AI: 7 Prompts That Catch Bad Numbers Before You Buy

A typical BRRRR underwriter spends 30-45 minutes per deal running ARV comps, rehab estimates, refi math, and cash-flow modeling. I know because I used to be that person — hunched over spreadsheets at 2am, second-guessing every line item. The 7 prompts below cut that to 8-10 minutes by letting an LLM do the structural analysis while you focus on the local-knowledge calls (neighborhood, contractor risk, exit liquidity). Each prompt is the result of testing against real deals — both wins and losses I've underwritten myself.

By Dale Weaver · Updated May 31, 2026. The framework below is the basis of Deal Flow — a 50-prompt pack for real estate investors using AI.

Why most investors underwrite the wrong number

The single most common BRRRR mistake isn't over-estimating ARV. It's under-estimating rehab. Specifically: forgetting the "invisible $8k" — the line items that aren't in the listing photos but always show up in execution:

• Permits + plan-review fees (county-dependent, $500-3500)
• Holding costs during refi seasoning period (3-6 months)
• Contractor change-orders (avg 12-18% of base bid)
• Utility transfers + temporary deposits
• Pest inspection findings (~30% of properties find something)
• HVAC sizing upgrade if you're adding sq ft
• Code-update items the inspector finds (smoke detectors, GFCI, etc.)

I learned this the hard way my first year. I'd run the numbers, they looked great — then the county inspector showed up and flagged the entire electrical panel. That was an unexpected $4,200. The first prompt in the framework forces a checklist run against your deal so these don't get missed at offer time.

The 7-prompt BRRRR underwriting framework

Prompt 1 — Invisible-cost scan

Paste the listing description + your initial rehab budget. The prompt walks through each invisible-cost category, asks you about each, and produces an adjusted rehab number. Catches $5-15k of missing budget on the typical deal.

Prompt 2 — ARV defense

Paste 3-5 comps you've pulled. The prompt cross-checks them on: square footage variance, days-on-market, sale-to-list ratio, and bed/bath mix. Flags comps that are likely outliers — estate sales, hoarder houses, contingent fall-throughs. Produces an ARV range with a defensible lower bound — the number you should underwrite to, not the optimistic mid-point.

Prompt 3 — Cash-flow stress test

Inputs: rent estimate, taxes, insurance, vacancy, capex reserve, property management, refi rate + term. The prompt runs three scenarios — base, rent down 10%, vacancy at 8% (vs typical 5%) — and tells you which scenarios still cash-flow positively. If only the base case cash-flows, the deal is too thin.

Prompt 4 — Refi seasoning math

Most BRRRR investors miss that the "seasoning" window (6 months minimum at most lenders) means you're holding the loan + rehab cost out-of-pocket for 6-9 months before recouping via cash-out refi. The prompt computes total cash-tied-up by month and tells you the maximum number of concurrent deals your bankroll supports. One student I mentored found he could only handle two deals at a time — not the five he'd planned. Saved him from a cash crunch that would've killed his portfolio.

Prompt 5 — Exit liquidity test

If the refi doesn't happen on schedule (appraisal low, rate market shifts, lender changes underwriting), your exit is selling the property as-is or doing a hard-money refi. The prompt asks about your local market: 30-day comps absorption rate, hard-money lender availability, and whether your contractor would do a partial rehab for a quick flip if needed. Three viable exits = deal is safe; one exit = too risky.

Prompt 6 — Tenant-quality projection

The neighborhood you buy in determines the tenant quality you'll attract. The prompt asks about local crime stats, median income, school ratings, and proximity to anchor employers — and projects probable tenant turnover rate, late-pay rate, and eviction probability. Buying in a market where the expected eviction rate is >15% requires very different management infrastructure than buying in a B+ market.

Prompt 7 — Final go/no-go scoring

Combines outputs from prompts 1-6 into a single 0-100 deal score. Above 70: pursue. 50-70: dig deeper, probably pass unless you have a specific reason to like it. Below 50: walk away.

What this framework won't do

The prompts handle the math. They don't handle the local knowledge that only matters for your specific market:

• Which contractors are reliable vs. which will ghost you mid-rehab
• Whether the neighborhood is gentrifying or stagnating (LLM has 12-month-stale data)
• Specific permit-office quirks in your county
• Whether the agent listing the property is dishonest about disclosed issues

As one seasoned investor I know puts it: "AI can tell you the square footage, but it can't tell you the smell of the basement." You still need to drive the neighborhood, walk the property, and know your contractors. The framework just makes sure the numbers are right so you're not making bad decisions on top of bad math.

Why prompts vs. a calculator?

Standard real-estate deal calculators give you a single number from a fixed formula. The prompts let you explain context the formula can't capture: "this is an estate sale so the seller wants quick close" or "the comp at 421 Oak sold at peak and won't replicate today." The LLM incorporates that nuance; the calculator can't.

Trade-off: prompts require an LLM (Claude, GPT-4, Gemini Pro) and 10 minutes of focus. Calculators are instant but miss the context. Use both — calculator for first-pass screening, prompts for deals worth real diligence.

Frequently Asked Questions

What LLMs do these prompts work with?

Vendor-neutral. Tested against Claude (Sonnet+), GPT-4+, and Gemini Pro. The structural prompts work the same on all three; the output formatting varies slightly but the underwriting math is consistent.

How does this compare to BiggerPockets calculators?

Different tool, complementary use. BiggerPockets gives you instant fixed-formula numbers. This framework lets you adjust the formula for your specific deal context (regional cost variation, unusual seller motivation, comp anomalies). Use BiggerPockets for screening 50 deals down to 10. Use this for the 3 you're seriously considering.

Does this cover multifamily / commercial?

The full 50-prompt pack includes multifamily-specific prompts (NOI project