Portfolio
What we’ve built, and what it changed
Every project here started the same way: with the people who would use it. A family’s game night. A historian tracing the forced migration of enslaved people into Texas. Hundreds of insurance agents competing for the same leads. An equipment lender’s evaluators, working from thousands of spreadsheets. Here’s what we built for each of them, and what happened next.
Our product
decks.app
From our family game night to thousands of users, with no advertising.
A real-time multiplayer game platform that turns the phones around a table into controllers. Simple enough for a 9-year-old and a 90-year-old, and good enough that players kept bringing their friends.
- Product design
- Real-time web app
- AI opponents
Client engagement
AI lead marketplace for a national insurance network
More than $2 million saved a year, and record-setting sales years.
Hundreds of independent agents were unknowingly bidding against each other for the same internet leads. We turned years of their lead receipts into an AI model that grades leads by quality, then built an internal marketplace where agents stopped competing with each other.
- AI lead scoring
- Data digitization
- Marketplace platform
Nonprofit partnership
Enslaved Texas
More than 50,000 freedmen in the handwritten 1867 voter registers, now on a map anyone can search.
A nonprofit research platform led by historian Deborah M. Liles, Ph.D., tracing the forced migration of enslaved people into Texas. We design and build the map, the search and the AI tools that read 160-year-old handwriting, with people checking the machines’ work against the original pages.
- AI handwriting transcription
- Interactive maps
- Research data
Client engagement
AI equipment valuation for an equipment finance company
Every machine valued twice, and growth that no longer waits on hiring evaluators.
Its know-how lived in thousands of one-of-a-kind spreadsheets, and the veteran evaluators who knew what every machine was worth were moving on. We had AI agents read a decade of workbooks into one platform, then built valuation agents that draw on those evaluators’ judgment and value every machine by two methods.
- AI valuation agents
- Data digitization
- Mobile field app
How we work
Software people love doesn’t happen by accident
A game, a history archive, an insurance marketplace and an equipment valuation platform have little in common, except the way they were built.
- 01
Start with the people
We learn who will use the software and what they’re trying to get done before we design anything. That’s why decks.app spread by word of mouth.
- 02
Use the right tools
AI where it pays off, whether that’s a language model, retrieval over your own documents or plain automation, and nothing where it doesn’t.
- 03
Keep improving
Automated testing and AI-assisted development let a small team ship changes in days, safely, long after launch.
Salt Branch Technology Group · Fort Worth, Texas
Let’s build software your people will love.
Tell us what it needs to do. We’ll tell you what it takes to build it right, including the parts that don’t need building.