SUPERHUMAN
REMOTE WORKER

01 / INSTALL THE SYSTEM

Up and running
In less than 30 minutes

A small silver mini PC with two front ports.
01

Mini PC

A small start with the complete system on one machine

A professional AI workstation tower with a dark mesh front and angular ventilation details.
02

Workstation

More room for agents, demanding work and local models

A full-height 40U rack of servers and switches.
03

HA cluster

Spread the system across nodes for capacity and resilience

STARTING POINT 8 vCPU · 16 GiB RAM · Kubernetes Allow additional capacity for agent workloads and local models

One chart
The whole system, ready

Choose your setup in the values generator. Install one Helm chart on your Kubernetes cluster, and bring the whole system online.

Your domain, services and license acceptance are set in values.yaml. Use the included services, or connect the ones you already run.

Generate values.yaml
srw — bash
$ helm install srw \
    oci://ghcr.io/superhuman-remote-worker/charts/superhuman-remote-worker \
    --namespace srw --create-namespace \
    --values values.yaml --wait
Pulled: ghcr.io/superhuman-remote-worker/charts/superhuman-remote-worker
NAME: srw   STATUS: deployed   REVISION: 1
$ kubectl -n srw get pods
NAME                                  READY   STATUS    AGE
srw-orchestrator-6d9c7-x2k4p          1/1     Running   41s
srw-orchestrator-6d9c7-p8m1q          1/1     Running   41s
srw-cockpit-58f4b-7hwzs               1/1     Running   41s
srw-postgres-0                        1/1     Running   58s
srw-postgres-vector-0                 1/1     Running   58s
srw-keycloak-0                        1/1     Running   55s
srw-gitea-0                           1/1     Running   55s
srw-nextcloud-0                       1/1     Running   52s
srw-agent-stateless-7c4d-kq2m         1/1     Running   30s
srw-agent-stateless-7c4d-z9f3         1/1     Running   30s

Scroll to see all columns

Cockpit: https://srw.your-company.example
First login: admin / see values.yaml

02 / SCALE THE AGENTSONE SYSTEM / EVERY KIND OF WORK

Made to fit
your demand

SRW separates its stateless agent harness from the workspace. Files, environments and saved progress live outside the agent process, so the fleet can scale like a regular deployment. One system serves your jobs, conversations and automations.

WORK COMES IN
  • Job API
  • Job MCP
  • Meeting transcription
  • Coding session
  • Website chatbot
THE CONTROL PLANE

SRW Orchestrator

Route work, manage lifecycles
and keep you informed

SHARED AGENT FLEET

Intelligence, on demand

Agent 1
Agent 2
Agent 3
Agent N

One shared fleet, from Agent 1 to Agent N, with each request bringing its own configuration

CONNECT TO YOUR SYSTEMSScoped connectors
  • External APIs
  • Internal APIs
  • Databases
  • Cloud folders
WORK IN THE RIGHT ENVIRONMENTDedicated workspaces
  • Virtual / S3
  • Containers
  • Virtual machines
Scale the fleet, keep the context Agents call connected tools and services, or execute code in a dedicated workspace. Workspaces and saved progress persist independently of the agent processing the request.
WORK FOLLOWS CAPACITY
Peak hours: priority work runs. Less urgent jobs keep their place and progress.

CAPACITY, PUT TO WORK

Priority work first
Progress all day

Demand changes throughout the day. SRW adapts how your fleet works to match it.

During peak hours, time-sensitive requests take priority. Less urgent jobs pause with their progress saved, freeing agents for the work that matters now.

When demand eases, those jobs resume where they left off. Research, reports and routine tasks fill the quieter hours, making useful work of the capacity you already have.

BUILT ON THE SAME FOUNDATION

Your infrastructure
Your entire agent fleet

Get notified Explore the agent configs