Seven components. Four layers. One control substrate.
LatentStack is an AI software stack an organisation can adopt one component at a time and run under its own model and data boundaries. Every component is available separately, and every component shares the same model-access and context policy.
What the agent knows.
Repository knowledge, team practice and the guidance that improves both. Agent-neutral: the context goes to whichever agent you run.
Which models see which work.
One policy substrate shared by every component: endpoints, model tiers, team permissions, and the restricted / permitted split.
Where the work happens.
Goal-based agent execution locally or in configured sandboxes, with execution kept separate from model hosting.
Who decided what.
The human layer: discussions, decisions, a feature wiki and agent sessions held on one topic.
Start with one layer. Add the next when the first has proved itself.
Choose a component. Agree deployment and model boundaries. Run a scoped POC against your own task set. Add further components when the first has proved itself.