Data & research
Deal Desk
A deal command center for multifamily real estate — from first sighting through years of ownership.
- Multifamily RE
- Underwriting
- Mock mode
The Bellvue Group
Don't take our word for it.
See it run.
Nine projects, built end to end — each one around a problem a business was actually stuck on.
Problems we've solved
What businesses bring us. What they leave with.
Barbershop
No idea what the shop really made — or what to hold back for taxes
Cuts in a notebook, money in a photo roll, taxes on a gut feeling. Now every cut, client and dollar lives in one place, with how much of each cut to set aside and the four dates estimated taxes are due.
Property inspection
Inspection notes retyped at a desk — and tools that die in the basement
Unit-by-unit walks used to mean paper, then retyping everything before it was useful. Now the inspector works on an iPad that saves with no signal, and the fund reviews the same property on a laptop with nobody retyping in between.
Real estate investment
Deals scattered across email threads and one-off spreadsheets
Rent rolls and T-12s arrive in every format imaginable. Now a fund drops them in, gets them underwritten, and keeps every deal in one place from first sighting through years of ownership — with AI costs kept under control by design.
Field sales
Prospects spread across a phone, a notebook and a string of texts
Nothing said who was hot, who had been called, or what came next. Now a rep has one lead desk — priority, stage and next step for every prospect — fast enough to update between stops and exportable in one click.
Real estate recruiting
Recruiting lists that were stale before anyone dialed them
Most lists fill their gaps by scraping and guessing. Now recruiting starts from newly licensed agents confirmed in state records, and every lead shows where it came from.
What we do
Three things, done properly.
Custom software for how you actually work
Deal Desk, DD Walk and Chaotic Kuts are each built around one job someone already does — not a generic template with someone else's workflow bolted on.
Data nobody else has, from sources anyone can check
Agentic Business Scraping and AdvisorSearch are built from public filings and directory records — SEC filings, state registrations, chamber directories — never guessed and never pulled from a source that says not to.
Agent crews that do the work at a cost that makes sense
Research and QA run on agent crews with their own subscription cost, not a metered bill that climbs with every step.
Featured work
Four of the nine. All of them running.
Field tools
DD Walk
A due-diligence inspection sheet an inspector carries through every unit — works in a basement with no signal.
- Offline-first
- 29 condition fields
- v0.4.0
Data & research
Agentic Business Scraping
A crew of agents that finds the business directories, checks what each one allows, pulls the listings and leaves a source on every field.
- 6 collection agents
- 56,655 businesses
- Source on every field
Data & research
AdvisorSearch
See the whole RIA succession market from public SEC filings — nationwide map to your shortlist in three steps.
- SEC filings
- 23,512 firms
- Provenance
The anchor engagement
Two of these
run a real investment firm.
Clayton Huck builds and runs this software for Westfork Capital Group. It's the longest-running client relationship at Bellvue, and the reason two of the nine projects on this site exist.
Deal Desk is Westfork Capital's deal command center. DD Walk is Westfork Capital Group's due-diligence inspection tool. Both are in active use and active development — Deal Desk has a live handoff document, and DD Walk is at v0.4.0 with phase 8 in progress.
In their own words
What people say.
"He is the smartest kid I've met, and I learn something new from him every day."
"The stuff he's done has had an impact on our firm."
"The stuff you're build is just incredible, man."
Process
How a build actually runs.
-
We work out what you actually need
A short spec before anyone builds.
-
We find out what's true
The research crew checks facts and sources, and a second agent re-verifies without seeing the first one's work.
-
We build it in parallel
Screens and logic at the same time, against one plan.
-
We try to break it
QA loops until it meets a written definition of done.
-
You get something that runs
Plus a plain-English summary of what it does.