Training
We teach your team to hold the tools themselves — human-directed AI development, practiced at your bench, on your real work. When the engagement ends, the capability stays. That's zero lock-in applied to people, not just code.
00 · the measure
Off-the-shelf fits nobody. We measure first.
Every engagement starts with how your people actually see the work. Then we build to fit — by hand, at studio speed, without the enterprise overhead tax.
01 · human-directed
Human-directed AI development
AI writes a lot of our code. It doesn't decide what to build, what matters, or when it's done — people do. Every line is directed, reviewed, and signed by a maker who understands your problem.
That's the whole trick. The machine compresses the timeline; the people keep the judgment. You get enterprise-grade outcomes at studio speed — without replacing anyone, least of all the people the work is for. Bespoke over templates, always: we measure first.
02 · the flywheel, expanded
We build the tools on our own ventures first. Then we hand you the proven ones.
The agentic delivery tooling, the reusable libraries, the identity and access work — each earned its place shipping ContentGuru, Hardshell, or TabbyTarot before it ever touched a client project. If a tool breaks, it breaks on our ventures, on our own risk. What graduates to your work is what survived.
Client work then funds the next venture, and the next venture forges the next tool. The loop is the business. ∎
03 · the foundations
Designed for portability from day one.
We start on Cloudflare's edge because it keeps your costs honest — and we build so you can leave. Clear migration paths to AWS, Azure, GCP, or your own racks. Identity and access handled from the first commit, not bolted on.
Your software is yours: export it, host it, hand it to another team. Zero lock-in isn't a discount line. It's a design constraint we cut against every day.
04 · the engagement
The shape of an engagement
No two engagements are identical, because no two shops are. What holds: we measure first, we ship in rapid feedback loops you can steer, and you stay in control — of scope, of priorities, and of the codebase itself. You see working software early and often, not a deck at the end of the quarter.
And the bench travels. Two more ways to work with us:
Public speaking
Talks and workshops on human-directed AI development, people-first software, and the studio-to-portfolio flywheel. Everything we present ran on our own ventures first — a working session from a working shop, not slideware.
05 · the diagnostic
Where most engagements begin: the QED Diagnostic
The Diagnostic produces a Business Cognitive Portrait — a living model of how your organization actually thinks, decides, and operates — interrogatable, embeddable, and working alongside you. Not a report you read once. A colleague that knows your business.
It ships with working agents built on the portrait, hosted by QED for 30 days. You use it before you decide anything.
The Diagnostic was $10K to start, $20K committed — the back $10K is now your choice:
Continue to Phase 2
$10,000 credit
The back $10K becomes a credit toward roadmap work. Portrait and agents stay live.
Keep the asset
$10,000 · you own it
Export and keep your portrait and agents — yours to run, embed, and keep. No obligation to hire us again.