Think of the PayCargo AI platform like a playground. The platform is the safe enclosed space. Each AI app is a piece of playground equipment. Some equipment is already installed (the chatbot slide, the observability seesaw). Some is on the way (the code-agents monkey bars). The playground itself never has to be rebuilt for a new ride to land.
The platform has five named parts. Every conversation we have about AI infrastructure maps onto one of them.
The safety features. Walls, gates, bouncy flooring, the supervisor watching the gate. What keeps kids safe regardless of which ride they climb on.
The installation hub. Where rides get mounted, where the operator manages access and connections to outside systems (Salesforce, GitHub, etc.).
The actual rides. Chatbot is a slide, observability is a seesaw, code-agents is the monkey bars. Each one is a self-contained piece of equipment kids can use.
Where new rides get designed before any metal gets bent. You sketch the slide, see what it would cost, check it for safety, then send the design to the Forge.
Where the rides are actually built from the Anvil's design. Our blacksmith is a coordinated team of AI models that turns a spec into running code.
Read top to bottom: a new app is sketched on the Anvil, built by the Forge, installed by the Console, and inherits every safety feature of the Foundation for free. No app gets to skip the Foundation. No app reinvents the Console's wiring. The whole platform compounds.
If you build a new playground for every new piece of equipment, you go bankrupt and the equipment is dangerous because the safety features are different every time. If you build the playground once and add equipment to it, equipment is cheap, safety is consistent, and everyone is faster. PayCargo is in the second category.
Each of the five docs in this series links to the relevant Step.* doctrine docs (Step 0 through Step 6.n) for the full engineering depth. Start with the cartridge or layer you're most interested in.