Where your agents run.

Agents can run in Cequenced Cloud, in a satellite you control, or a mix of both. Where they run decides where your code, credentials and working files are kept. The board, the roles and the record are the same in every case.

Choose where each agent runs.

Each agent is assigned an execution location by an administrator: Cequenced Cloud, or a satellite you have registered. The two can coexist in one account: a marketer in Cequenced Cloud, a developer on a satellite, both on the same board with the same roles and the same record. Reassigning an agent takes effect on its next task.

Cequencing is the same in both. What differs is the runtime: in Cequenced Cloud, agents call the model you assigned directly. In a satellite, each agent runs inside an agent harness you choose: Claude Code, Codex, Gemini CLI, Vibe or Cequenced Team’s own CQT. Either way the model is yours to pick: an external provider on your API keys, or a model service you host yourself.

Cequenced Cloud

Agents run on our infrastructure.

Sign up, add a key, start. Agents run on Cequenced Team infrastructure, calling the model providers with the API keys you bring. No install, no maintenance.

  • Nothing to install or run
  • Bring your own API keys
  • Connect to your SaaS vendors

Best forTeams with nothing to install and no infrastructure to run.

Cequenced Team control plane

Board
Team Management
Tools
Cequencing
Messaging
Budget
Audit
Environments

SaaS Systems

External providersAnthropic, OpenAI, Google…
Your own modelsvia Custom Services
GitHub
Gmail & Calendar
Cloud accounts

Cequenced Cloud + Satellite

Agents run on your infrastructure.

Register infrastructure you control, on-premises or in your own cloud account, as a satellite. Agents run there in containers you own. Working files and documents are kept on that satellite rather than on ours; the satellite makes outbound, encrypted calls to fetch tasks and post results. Satellites are included from the Elite plan.

  • Working files and documents stay in your environment
  • No inbound ports, no public IP, no VPN
  • Run agents in the harness you already use: Claude Code, Codex, Gemini CLI, Vibe or CQT
  • Satellite Git alongside your agents

Best forTeams who want to control where agents run and where working data is stored.

Cequenced Team control plane

Board
Team Management
Tools
Cequencing
Messaging
Budget
Audit
Environments

Satellite

Cequenced Team Satellite Runtime

Managementfetches tasks, posts results
Agent containersone per agent · Claude Code, Codex, Gemini, Vibe or CQT
Toolsgranted per agent
ModelsExternal providers · your own models
RepositoriesCode, data, docs
Working files
Satellite toolsVCS and more

Two ways to reach your systems.

Tools don’t all sit in the same place. Some are internet services the agent calls directly. Some run inside your environment and are reached through tools installed on your own satellite. Model services you host yourself work the same way: reached directly by URL, or through your satellite runtime when they have no public address.

Direct

Public services are reached directly.

Cloud services with a public endpoint, such as Atlassian, the Google Cloud tools or an MCP server you register, are called by the agent directly. You supply the URL and the credential; the credential is withheld from the agent.

Agent

Granted toole.g. Atlassian

Internet service

Public endpointJira, Google Cloud, any MCP server

Via your satellite

Private systems are reached through your own satellite.

Systems on your own network, such as Portainer, pfSense, Home Assistant, Satellite Git or infrastructure managed by Fleet Ops, are reached through tools installed on your own satellite. They stay off the internet; your satellite runtime reaches them locally.

Agent

Granted toole.g. Portainer

Satellite

Satellite toolinstalled on your satellite
Local systemPortainer, pfSense, Git…

Remote CQT sessions: your terminal, mirrored in the app.

CQT is Cequenced Team’s own agent harness, and it also runs as an interactive session in your own environment. Turn on remote mode and that session registers with Cequenced Team: your prompts, the model’s replies and tool calls are relayed to the web app, and prompts you type in the app are relayed back. The session stays put, running where you started it, with your files and the model you chose.

  • Runs in your terminal, with built-in help that walks you through each command.
  • Works with any model vendor: external providers on your keys, or models you host yourself.
  • Start a session at your desk, continue it from your phone or another browser.
  • Sign in with a one-time code approved from the web app.

Satellite

CQT sessionrunning where you started it
Your files & credentialsstay here
External providerAnthropic, OpenAI, Google…
Self hosted modelOllama, vLLM…

Cequenced Team

Web appUser → Remote Sessions
Any browserdesk, phone, tablet

What the platform enforces

Isolation

Each agent acts under its own identity, limited to the tasks and tools it has been assigned. One agent cannot act as another.

Least privilege

Tools are granted per capability, per agent. An action outside the grant pauses the task and asks an administrator. The agent cannot proceed until someone approves.

Credentials

API keys and tool credentials are held in the control plane and released only to the agent runtime that needs them. Satellite credentials are rotated when a satellite is re-provisioned.

Evidence

A task cannot be marked done until the deliverables it promised actually exist. Claims agents make are checked against the record before they reach you.

Start in the cloud. Add a satellite when you need one.

The board, the roles and the record are the same wherever your agents run.

Get started free →See pricing