Real software, real screens, and no numbers we haven't earned.
Client work appears here with permission and with the client's name on it. Everything built for ourselves
is labeled as ours. No invented testimonials, no borrowed logos, and no results claimed that haven't
actually been measured.
Recent projects
Real screens from real software.
Every entry is labeled for what it is, and nothing is claimed that hasn't been measured.
Client projectAI speech tool
TalkTalk Global — accent clarity trainer
The practice tool that replaced marking pronunciation homework by hand.
Two speech-language pathologists built a real accent clarity program, then hit the ceiling every practitioner hits: every student recording had to be listened to and marked up personally. We built the tool that scales that. Students work through twenty standard clinical passages — the Rainbow Passage, Arthur the Rat, Comma Gets a Cure — chosen because between them they cover nearly every sound in English. Each one is read aloud, recorded in the browser, played back, and submitted for scoring on fluency, accuracy, and pronunciation. Completed passages are checked off and kept in a running history, so a student has a real baseline and can watch it move without waiting on a human review cycle.
In-browser audio captureTwenty clinical assessment passagesFluency, accuracy & pronunciation scoringPlayback and re-record before submittingProgress tracked across passagesEmbeds inside their Kajabi course
An operations platform for an industry running on spreadsheets.
Live sports card breakers price breaks, track inventory and expenses, publish break lists, run overlays during a stream, and reconcile results afterward — nearly all of it repeated from scratch in spreadsheets for every single show. The first working version was built in a seven day sprint, then extended over the following month into one system covering the entire lifecycle of a break, with the real pricing logic encoded and profit visible before, during, and after.
The marketing site that turns a cold visitor into a free trial.
A conversion-focused site built to sell the platform: a single clear offer, a free break calculator as the lead magnet, and a trial signup as the one action every page points at. Built for speed and search, with the pricing, FAQ, and proof a buyer needs before committing.
Static site, no build stepFree calculator lead magnetConversion-focused copySEO + structured dataSub-second load times
An AI that takes real actions, not one that just talks about them.
A personal AI operating system built to run daily business operations end to end: it searches and reasons over an indexed notes vault, drives a real Chrome browser to look things up or take action, controls other Mac apps directly, and runs a live day-by-day operations board — surfacing exactly what needs doing today, drafting the outreach, and flagging what still needs a human "go." Every tool call is shown live as it happens rather than hidden behind a spinner. Built to prove what an AI agent can actually do inside a real business, not a demo of what a chatbot can say.
Agentic tool-use loop (Claude)Live Chrome control via CDPCross-app macOS automationVector-indexed notes vaultVoice interfaceReal-time activity feed
Capability
Everything The 7-Day Sprint touches.
The list below is what goes into a build week. Not every project uses all of it. The scope document says which parts yours does.
Product and workflow
Process mapping, workflow architecture, user flows, data modeling, and deciding what deliberately doesn't get built.
Interface
Interface and experience design, responsive desktop and mobile layouts, and the states that real use produces rather than only the happy path.
Application
Front end and back end development, database setup, authentication and access control, and business logic that matches how you actually price and operate.
Integrations
Third party APIs, connections to the systems you already run on, automated data movement, and secure configuration of credentials you provide.
AI where it earns it
Summarizing, extracting, classifying, drafting, and assisted workflows. Used to delete work, not to have something to put on a slide.
Delivery
Functional QA against the written scope, deployment to infrastructure you own, training, documentation, and full ownership transfer.
Transparency
How the first client case studies will be documented.
Every sprint client is asked, after delivery and only with written permission, for the same set of things.
This is the format every case study on this site will follow, so you can judge them properly when they exist.
The original workflow and what it was costing, in hours and dollars
The tools and workarounds it was running on before
What was built, with screenshots or a short demo
Time from kickoff to delivery
Hours saved per week, measured rather than estimated
Errors or rework reduced
Revenue gained or protected, where it can be attributed honestly
A quote from the person who uses it daily
What was still true thirty days later
If a result can't be verified, it doesn't go on the page. That rule costs us conversions in the short term
and it is the only version of this we're willing to run.
Founding sprint availability is open.
We're accepting a limited number of sprints each month while we build this first portfolio of documented results. That's a good moment to be on the other side of the table.
