Robert Wensley turned this week's call into a hands-on LiftGPT workshop: a short lesson on how to think about underwriting, followed by two live deals analyzed with attendees and open Q&A.
The mindset shift: underwriting isn't about picking one number. Robert's point is that ARV is a spectrum of outcomes that depends on what you do with the property, and the biggest assignment fees come from spotting the highest-value use, not just comping the house as it sits. Examples covered include knocking down and rebuilding on an existing foundation, adding a pop-top or basement, and adding an ADU.
What we covered:
Why ARV should be treated as a range, and how LiftGPT shows a distribution curve instead of a single number
Why the headline ARV is set at the 70th percentile of that curve: fixed-up properties typically sell about 5-20% above the median price per square foot
Why older comps can still be useful if you time-shift them for market movement, and how LiftGPT looks at every nearby transaction, adjusted for beds, baths, square footage, and sale date, rather than a handful of comps
How price point affects wholesaling economics: higher price points mean higher cost per lead and lower lead-to-deal conversion, while lower price points mean smaller fees for the same amount of work
Live deal #1 (San Antonio, TX): a subject-to property with a 2.6% loan, analyzed across selling, renting, lease-to-own, and a wrap with owner financing - the wrap came out as the strongest exit
Live deal #2 (Palm Coast, FL): a cosmetic flip analyzed with an ARV range, rehab estimate, max offer at 12% and 2 points, rental comps, market trends, and permit history - including why the seller's ARV looked too high
Tips for getting more out of LiftGPT:
Tell it your real numbers (cost of capital, points, vacancy) and ask it to save them to memory, so every later analysis uses your assumptions
Ask for short answers if the full analysis is too long - for example, "give me the cliff notes of your last response"
Upload inspection reports or contracts directly into the chat so the analysis reflects the actual scope of work
Use the ghost icon next to the submit button for an incognito chat that isn't used for training
If a result looks off, give it a thumbs down and use the feedback button - the team reviews these to improve the tool
Coming up, as discussed on the call: selecting specific comps on the map and having the numbers recalculate, highlighting as-is purchases versus fixed-up flips on the distribution curve, deeper scenario analysis, scheduled searches across Investorlift inventory, and a connection with the new God Mode.