“Checkout button does nothing on my iPhone??”
✓ FixedRepro 3/3iOS Safari
AI-native studio · Software · Systems · Experiments
If you can describe it,
we can probably
build it.
Just tell us, simply and plainly, what you need. No jargon, no forty‑page brief. We’ll work it out with you, then build it properly.
New VibeHelm · our multi-agent workspace is in early beta02New · Early beta open
Your multi-agent workspace.
Run a crew of AI coding agents side by side on Windows 11: Claude Code, Codex, OpenCode, Antigravity, Cline and MiniMax Code, on your own plans and keys. Give them one job together, race them, benchmark them and record the lot.
- Vibe CodeAll your projects, all at once
- CollabOne crew, one project, planned together
- Face OffAgents race the same job, side by side
- BenchmarkSet challenges, every number measured
- StudioRecord any mode for your videos
03The studio
One material. Many forms.
Every idea arrives a little molten. We give it shape: discovery, prototype, test, ship. Some of what we cast are studio tools, some are standalone products, and one is a suspiciously well-informed butler.
04Discovery · AI-guided
Start here. Oliver knows the way.
Oliver turns early-stage ideas, half-formed requirements and “I know what I want, I just can’t explain it” into structured direction a real project can move on.
- Guided discovery without the corporate questionnaire
- Plain-English questions, one at a time
- You leave with a brief, not a sales call
OliverDiscovery guide
listening
- I run a small bakery. I want an app… or something? People keep asking about pre-orders.
- Good place to start. When people ask about pre-orders, what are they really after: reserving something, or skipping the queue?
- Both, honestly. Saturdays are chaos.
- Then version one probably isn’t an app. Let me draft a brief you can react to.
BriefDraft 01
- Problem
- Saturday queues and lost pre-orders
- First version
- Order-ahead page with collection slots
- Not yet
- Native app, loyalty scheme
- Next question
- Which ten products sell out first?
Illustrative conversation
05AI systems readiness
Ready for AI? Find out before you spend.
A short, guided check of the things that decide whether AI actually helps a business: its data, its processes, its people and its tools. You come away with a clear readiness score and the few things worth fixing first.
Readiness report
Sample
0/100
Ready to pilot
One workflow, with guard rails
Fix first: write down how the work actually flows, and agree simple rules for using AI with customer data.
06Quality · Release confidence
Release Gate. Ship on evidence, not hope.
A practical QA and release layer for finding bugs, understanding severity and turning messy feedback into a clear route to ship.
- 01Triage
- 02Evidence
- 03Repair flow
- 04Ship
In development
Gate
Gathering feedback
Hold · 2 blockers
Verifying fixes
Clear to ship
Incoming feedback
Blocker
Major
Minor
“Can’t log back in after resetting my password”
✓ FixedRepro 2/3Auth log
“Total doesn’t update when I change the quantity”
Repro 3/3Basket
“Logo looks a bit blurry on my laptop”
1 reportRetina
Illustrative example
07Coming soon · Cecil
Cecil requests the pleasure of your company at
A Nightcap at Ravensmere
Sir Edmund rang for his port at eleven. By midnight he was dead.
An AI-hosted murder mystery for one table of friends. The big screen tells the story. Your phone tells yours. And Cecil, the butler, knows every secret in the room.
- 3–4 players
- About 15 minutes
- One screen, everyone’s phones
- No human game master
“I am no physician, but I have read a great many novels.”
Cecil
08Also in the foundry
A single-player space trading game. Trade, mine and race five rivals to the first million in a living star system.
In development · Beta testing coming soon
orbitalfortune.com
09DF9 Labs · Independent AI research
Intelligence. Under observation.
DF9 Labs is our independent AI publication. We build, test and investigate artificial intelligence, then share what the evidence shows.
- The Model Gauntlet: AI models take on the same real builds, judged on technical evidence and by a human
- Research such as Civilisation One, a society of AI agents under observation
- A journal and AI news, written from what we actually saw
10How we work
Human judgement. Machine speed.
We use AI aggressively where it helps, and keep people firmly in charge of the decisions that matter. The aim isn’t “more AI”. It’s less friction, better software and more room for good ideas.
We prototype assistants, internal systems and lightweight software for small teams who need useful automation without turning their business into a science project.
We cast
- Product design
- AI workflows
- Rapid prototypes
- QA systems
- Creative technology
- Assistants
- Internal systems
- Games
- 01Discovery
- 02Prototype
- 03Test
- 04Ship
da·tum noun — a fixed point from which everything else is measured.
Small studio. Serious curiosity.
Bring us the fuzzy version. That’s usually where the good stuff is.