Visca
The Stack/Lifecycle plane

Runtime

Lattice Runtime

The open-source runtime layer — and the foundation the whole stack runs on.

The runtime layer of the stack, and its open-source foundation (MPL 2.0). Runs autonomous workloads across cloud, edge, on-device, and on-robot — with first-class awareness of tokens, models, memory, and embodiment, routing over open-weight models so cognition stays in your walls.

Lattice Runtime is a primitive of the Visca stack — it ships with the stack, versions with the stack, and audits with the stack. It is not separately purchasable.

What it is

The role Lattice Runtime plays in the stack.

Lattice Runtime is where autonomous actors execute, on compute you control — and it is the open-source foundation the rest of the stack is built on. It is the wedge: anyone can read it, fork it, and run the full substrate inside their own walls under MPL 2.0, with no accounts, no telemetry, and nothing calling home. It schedules, runs, and supervises actors across heterogeneous compute with awareness of the dimensions that distinguish autonomous workloads from generic ones: token budgets, model availability, tool dependencies, memory lifecycle, latency objectives, embodiment constraints. Every actor runs under its Sigil, a budget, a model routing policy, a memory binding, and a safety envelope — and every execution is recorded to Chronicle. Model routing favors open weights in-perimeter, so inference need not leave your boundary. Built on durable execution: when a process crashes, work resumes from the last checkpoint — state is never lost.

The problem

What goes wrong without it.

Generic schedulers are blind to the dimensions that determine whether an autonomous workload succeeds. Kubernetes does not understand token budgets. Lambda does not understand provider fallback. Robotics middleware does not understand cost ceilings. Operators are left running autonomous workloads on infrastructure designed for stateless requests, with autonomy-specific concerns built into application code, inconsistently, across every team — and routed out to hosted model APIs that send regulated data straight past the perimeter.

Capabilities

What Lattice Runtime does.

Open and commercial

Built in the open. Run in your tenancy.

Lattice Runtime has two surfaces: an open-source reference in Lattice Runtime (MPL 2.0), and a managed delivery as part of the Visca stack on Visca Cloud. Features in the open never move behind the paywall.

Open

Lattice Runtime — open foundation

MPL 2.0 · self-hostable · foundation-track governance

  • ·Single-node reference runtime
  • ·Local executor
  • ·Model routing primitives
  • ·Memory adapters (in-memory, local disk, SQL)
  • ·Behavioral envelope enforcement

Commercial

Visca Cloud — managed sovereign delivery

Your tenancy · dedicated · self-managed · air-gapped

  • ·Multi-cluster control plane
  • ·Hosted Lattice Runtime executors across regions
  • ·Provider rate-limit pooling across customers
  • ·Cost governance across organization
  • ·High-availability control plane
  • ·Compliance-grade isolation tenancy

LIFECYCLE · same plane

Inside the lifecycle plane with Lattice Runtime

Across the stack

Composing with the other planes.

The whole stack. Self-hosted. One ecosystem.

The entire agent stack, inside your own walls.

Models, identity, tools, voice, payments, runtime, and audit — as one integrated ecosystem, self-hosted, sovereign, air-gapped. Nothing stitched from vendors. Nothing leaves your perimeter. Open at the core. No license rug-pulls, ever.