Operum is a desktop application that orchestrates 6 specialized AI agents (PM, Architect, Engineer, Tester, Marketing, Community) to automate your development, marketing, and community workflows.

Is my code safe with Operum?

Yes, your API keys and secrets are stored securely in your OS keychain and never leave your machine. Operum doesn't store or collect your code - AI calls go directly from your machine to the AI provider (Anthropic).

What platforms does Operum support?

Operum supports macOS (Apple Silicon and Intel), Windows, and Linux (Debian, Ubuntu, Fedora, Arch).

How much does Operum cost?

Operum is free during the open beta period. No credit card required. Paid plans (starting at $49/mo) begin when the beta ends.

Your First Session with Operum

This guide walks you through exactly what to expect from the moment you download Operum to your first successful agent task. No surprises - just a clear path to AI-powered automation.

Download & Install

~2 minutes

Download Operum from operum.ai/download. The installer is available for:

macOS - Apple Silicon (.dmg) or Intel (.dmg)
Windows - Installer (.exe)
Linux - Debian (.deb), Fedora (.rpm), or Arch (convert .deb)

Run the installer and follow the prompts. Operum will install to your Applications folder (macOS), Program Files (Windows), or system path (Linux).

The login screen you'll see on first launch

Install Dependencies

~3 minutes

On first launch, Operum checks for required tools. If any are missing, you'll see a dependency installer:

Git Version control for your code

GitHub CLI (gh) For GitHub integration and authentication

Claude Code The AI engine that powers all agents

macOS/Windows: Click the install button for each - it's automatic via Homebrew or winget.

Linux: Copy the displayed commands to your terminal (requires sudo).

Tip: If you already have these tools installed, Operum will detect them and skip this step.

Sign In

~1 minute

Click "Login with Operum" to authenticate:

Your browser opens to operum.ai
Sign in with GitHub or Google OAuth
You're automatically redirected back to the app

Your session is stored securely with automatic token refresh - you won't need to log in again unless you explicitly sign out.

Create Your First Team

~2 minutes

A team is a collection of AI agents working on a specific project. Each team connects to a git repository.

Click "Create Team"
Enter a name (e.g., "My SaaS Project")
Select a git repository directory on your machine
Click Create

Operum creates a .operum/knowledge/ folder in your repo to store project context. This is git-versioned, so your knowledge base is backed up with your code.

Your dashboard after creating a team - ready to start agents

Meet Your Agents

Understanding the team

You have 6 specialized AI agents, each with a specific role:

PM Agent

The coordinator. Manages priorities, tracks progress, and keeps everyone aligned. Start here for planning.

Architect Agent

Reviews tasks for technical implications. Provides architectural guidance and maintains design docs.

Engineer Agent

The builder. Implements features, fixes bugs, writes tests, and creates pull requests.

Tester Agent

Quality assurance. Reviews PRs, runs tests, reports results, and approves for merge.

Marketing Agent

Growth focused. Handles SEO, content strategy, social media, and analytics.

Community Agent

User support. Monitors Discord, Twitter, Reddit. Responds to questions and feedback.

The Agents page showing all 6 specialized agents

Start an Agent

~30 seconds

Starting an agent is simple:

Go to the Agents page or use the dashboard
Click Start on any agent card
The agent opens in a terminal window powered by Claude Code
The status indicator turns green when ready

Recommendation: Start with the PM Agent first. It can help you plan tasks and coordinate with other agents.

You can start all agents at once using the Start All button, or enable Auto-start in Settings to launch them automatically when Operum opens.

Give Your First Task

~1 minute

With an agent running, you can give it tasks. From the dashboard, use the command input:

Review the codebase and suggest 3 improvements we should prioritize this week.

Or for engineering work:

Fix the login form validation bug mentioned in issue #42.

The agent will:

Analyze your repository (if connected)
Access the knowledge base for project context
Execute the task based on its specialization
Report results in the Activity Feed

The dashboard showing agents working with real-time activity

Monitor Progress

Ongoing

Operum provides several ways to track what's happening:

Activity Feed

Real-time stream of all agent actions, completions, and requests.

Pipeline Board

Visual Kanban-style board showing issues moving through stages.

Metrics Dashboard

Token usage, PRs merged, issues closed, and productivity stats.

Status Bar

Quick glance at how many agents are working, idle, or need attention.

The pipeline board tracking work across development stages

Build Your Knowledge Base

Over time

The knowledge base is shared context that all agents can access. It includes:

Project info - Name, description, website, repository
Goals - Strategic objectives and near-term priorities
Architecture - Technical documentation and design decisions
Business - Business model, processes, and workflows
Links - Important URLs, social media, documentation

The PM agent can help you build this through a guided onboarding conversation, or you can edit files directly in .operum/knowledge/.

The knowledge base editor with project context

Let Them Work

Autonomous operation

Once configured, Operum agents work autonomously:

PM monitors other agents and pushes work forward
Engineer creates PRs for completed tasks
Tester reviews PRs and runs tests automatically
You review and merge when ready

You'll receive notifications when agents need your input (approvals, decisions, clarifications). Otherwise, they work in the background while you focus on what matters.

Pro tip: Check the Activity Feed at the start and end of your day to see what was accomplished and what needs attention.

What to Expect Day-to-Day

Agent Responses

Agents communicate through status prefixes:

DONE: - Task completed successfully
REQUEST: - Agent needs your input or approval
ERROR: - Something went wrong (usually recoverable)

Automatic Coordination

When one agent completes work, the PM is notified and can delegate follow-up tasks. For example: Engineer finishes a PR → PM tells Tester to review it.

GitHub Integration

With GitHub connected, agents can:

Read issues and PRs for context
Create branches and commits
Open pull requests with descriptions
Update issue labels based on progress

Token Usage

Agents use Claude API tokens. Monitor usage in the Metrics dashboard. Costs depend on task complexity - simple queries use fewer tokens than large code changes.

Common Questions

Can I use Operum without GitHub?

Yes. GitHub integration is optional. Agents can work with local code, though some features (automatic PR creation, issue tracking) require GitHub.

What happens if an agent gets stuck?

The PM agent monitors for stuck agents and alerts you. You can also manually stop and restart any agent, or give it clarifying instructions.

Can I work on code while agents are running?

Yes. Agents work in isolated environments. However, avoid editing the same files an agent is working on to prevent conflicts.

How do I stop all agents?

Click "Stop All" on the dashboard, or close the Operum app. Agents stop gracefully and save their state.

Ready to get started?

Download Operum and have your AI team running in under 10 minutes.

Download Operum