Praecise
Menu
CloudPlatform · Hardware we own and operate

The Cloud.

The managed side of the same app. Apps, agents, models, data, machines and automations, provisioned through the platform and recorded by it: a deploy is a decision with a ledger entry, not a file somebody placed.

Runs
Apps · Agents · Models · Data · Machines · Automations
Source
A folder, a container, a static export, or a git push
Edge
Certificates issued, traffic routed, releases kept for rollback
Hardware
Our own machines, provisioned by region as a workload needs, including NVIDIA Grace Blackwell class GPU nodes
Terms
A commercial service; the Lab stays open source
Access
Early workloads placed by conversation
Open the ConsoleRequest access

01

What it runs(6)

Six nouns, each with one way in. The same names the Harness uses, so an app never learns a second vocabulary.
  1. 01

    Apps

    A folder, a container or a static export.

    Built on our machines from your source, placed behind our edge with a certificate issued, and supervised. Node, Python and static sites are recognised from the tree; nothing to describe twice.

  2. 02

    Agents

    Agents as a service.

    An agent from the Harness is served as a chat UI, a REST endpoint, an MCP server and an A2A card. Run yours, or ask us to run one for a job you name.

  3. 03

    Models

    A ladder, not a model id.

    Fast, balanced, best. The models behind each rung are ours to keep current, accelerated by Praecise Engine on our hardware. Point the same app at your own endpoints instead and nothing else changes.

  4. 04

    Data

    Postgres, vectors, time series, key-value, objects.

    A tenant is provisioned with one command and the app is handed its credentials. Vector and time-series extensions where the licence allows them; object storage for what does not belong in a row.

  5. 05

    Machines

    Virtual machines and GPU compute.

    For workloads that need a whole box: a VM with a fixed shape, or a GPU node in the class the Engine is tuned for. Hardware we own and operate, provisioned in the region a workload needs.

  6. 06

    Automations

    Workflows that wait.

    Scheduled or event-driven runs built from the Harness's durable workflows: they pause for a person, resume where they stopped, and leave a journal that says what happened.

02

How a deploy works

Three verbs. The tier is read from the tree; the release is built on our machines and kept.

01 · Describe

A tree that says what it is

A Node app with a build script, a Python service, a static export, or a Harness folder. Secrets are set on the site, never committed. Nothing is described twice.

02 · Place

One command, or a push

The source is built into a release on our machines, given a port and a unit, and put behind the edge under its names. The previous release stays where it is until the new one answers.

03 · Run

Supervised, observed, reversible

The process is supervised, its health is watched by name, and the bytes served can be checked against the ones approved. Rolling back is pointing at the release that was there before.

$ praecise deploy ./my-app my-app my-app.example.com
  detected   node · next (static export)
  built      release 7c1e…
  placed     https://my-app.example.com
  verified   bytes served = bytes approved

An agent is a deploy like any other. The folder that ran on a laptop with praecise dev is placed the same way and comes up serving a chat UI, REST, MCP and an A2A card under its own name.

Data is provisioned the same way: one command creates the tenant, and the app is handed its credentials through the site’s environment.

03

Bring your own, or bring one key

The framework is open source and complete on its own. The Cloud shortens the config; it never changes the code.

Your own endpoints

// praecise.config.ts
export default defineConfig({
  models: {
    house: {
      url: "https://models.internal",
      credential: "HOUSE_KEY",
      fast: "…", balanced: "…", best: "…",
      room: 200_000,
    },
  },
});

The first endpoint whose credential is present is the one that runs, and a declared endpoint always beats the Cloud.

One key

# .env
PRAECISE_API_KEY=…

The models behind the ladder, the database and object storage are provisioned for you and reachable through the same code. There is nothing to choose and no model id to keep current.

04

Machines

For work that needs a whole box, or a GPU node in the class the Engine is tuned for.

Inference on the Cloud runs through Praecise Engine on hardware we operate. What is measured in the Lab is what serves here: speculation only where it pays, licence-checked drafters, admission that knows the output length, and a decode path tuned for the memory-bandwidth ceiling of the machine it runs on.

Virtual machines with a fixed shape, and GPU compute for training, batch jobs and models that do not fit the ladder. Machines are placed the way everything else is: through the platform, with a record.

GPU class
NVIDIA Grace Blackwell (GB10), where the Engine is verified
CUDA
One build, Turing through Blackwell
Shapes
VMs with fixed CPU, memory and disk; whole GPU nodes
Placement
Through the platform, recorded in its ledger

05

Sign in, or ask.

The Console is where a project is deployed, configured and watched: sign in with your address and a code. If your address has no projects yet, describe what you are running and we will reply with what it takes to place it.

console.praecise.comcontact@praecise.com