Product note
One foundation, several Decks
Autodeck keeps multiple Decks on one shared foundation so products can differ without fragmenting identity, runtime governance, or deployment.
One foundation, several Decks
Autodeck is not trying to become one giant application with an AI feature bolted onto every corner.
It is also not trying to turn every new product idea into a separate mini-stack with its own identity model, runtime decisions, and deployment story.
The product sits between those two mistakes.
Why several Decks exist on one foundation
Different kinds of work need different interfaces, workflows, and agent behaviour.
That is why Autodeck already includes distinct surfaces such as:
- Cloud Analyzer Deck
- ArchDoc Deck
- Toetsie Deck
- Dungeon Master Deck
These are not cosmetic variations of the same page. They are meaningfully different working surfaces.
What they still share
Even when the products differ, they should not each reinvent the operating model around them.
Keeping Decks on one foundation allows them to share:
- workspace identity
- tenant boundaries
- governed agent execution
- common infrastructure and deployment patterns
- consistent navigation and workspace controls
That gives Autodeck a stable spine while still allowing each Deck to behave like its own product.
Why “Deck” is more useful than “app”
The public term matters here.
"App" usually tells you very little. It suggests a screen or a feature area, but not how the thing actually works.
"Deck" is more useful because it implies a composed unit:
- interface
- model choices
- tools
- agent behaviour
- domain intent
Internally, the codebase may still talk about app packages. Publicly, Deck is the better product term because it describes what a user is actually stepping into.
The architectural reason
Autodeck is deliberately built so Decks can differ without turning the whole system into a mess.
That means:
- Deck-specific behaviour stays local to the domain
- the shared runtime remains shared
- boundaries stay explicit enough for future extraction if it is ever justified
This is a more disciplined way to build AI products than either extreme: one shapeless mega-product or a sprawl of disconnected tools.