Duplocloud

The DevOps Agent Platform

Build any platform engineering use case in hours.

See Demo Run it on your laptop →

Describe the use case

— tell the coding assistant what you need, in plain English.

Assistant builds the agent

— leverages what the platform provides; builds only your custom remainder.

Test locally, push

— ready for your end users to consume.

Hundreds of organizations trust Duplo Agents with their cloud infrastructure

Why duplo

Scale DevOps without scaling headcount

Clear the backlog

Agents absorb routine ops work. Your engineers keep the judgment calls.

Developer self-service

Devs ship without filing tickets — guardrailed by workspaces and RBAC.

50%+

Lower token costs

Shared sessions, model router, token policy, and token-less workflows.

Single pane of glass

Every agent, workflow, and token — visible, audited, in one place.

Compliance built in

SOC 2, HIPAA, PCI — RBAC, secrets, audit trails, human-in-the-loop by default.

No model lock-in

Claude, OpenAI, Gemini, open-source — swap per workflow, administratively.

1 · Multi-tenancy

SSO
Workspaces
Token quotas
User management
RBAC & policies

2 · Multi-player AI

Shared N-player Claude sessions
Skill distribution
Context & agent memory
Knowledge base & vector DB
Token management policies
Model router

3 · UX

Standardization, not vibe-code
Reusable repeat-function modules
Slack/Teams · API · CLI · MCP

4 · Security

Secret management
Credential management
Human-in-the-loop
Command approval policies
Audit trail

5 · Observability

Agent activity
Token usage
Skill usage
User activity
Tool calls

6 · Cloud Capabilities

Kubernetes management
TF management
Deployment orchestration
Pipeline management
Security best practices
Compliance controls
Multi-cloud services (S3, RDS, GCS, Azure SQL and 100s of others)

Build Your Own

Any use case becomes an agent

Each use case is about provisioning, updating and deleting a set of resources. Each resource has an input form, status display and a set of actions associated. The user runs the platform locally and interacts with a coding assistant.

Define a policy model — a taxonomy of the resources your workflow manages. Each typed spec becomes a form and an API contract automatically.

Network Cluster Environment Namespace Workloads

Lifecycle — describe how the resource should be provisioned, updated and deleted. Point to any existing automation or tools that should be leveraged.

Generate package — the framework generates an agent package: UI, a skill to manage the lifecycle, and APIs for actions on the resource once provisioned.

Test locally — the framework updates the local copy of the platform with the agent, which the engineer tests via the localhost URL.

Ship it — publish the agent to Git and deploy to the hosted version of the platform in your org.

Watch: an engineer builds and tests an agent end-to-end on a laptop, then publishes it to their org — in under an hour. Open on Vimeo ↗ · Deep dive: The Agent Framework →

Extension Studio · DevKit

Build extensions from your laptop

The DevKit gives extension developers the whole platform locally — one Docker Compose, an example extension, a Claude skill to build your own. Think Replit for DevOps.

Download & run — the full platform boots locally.

Build & test — develop your extension end-to-end with Claude.

Publish — push it to your org’s portal. RBAC, audit, and cost controls attach automatically.

The platform itself runs self-hosted on your Kubernetes — or fully managed by DuploCloud. The laptop is just where extensions get built.