AgentDock

Your AI.Your tools.Your entire workflow.

Bring your AI subscriptions together in one desktop app. Build websites, develop software, automate work and take projects from idea to completion.

Point an agent at a folder, describe the site, review each change before it lands, and keep going until it ships.

Try asking for: A five-page site for a local business: layout, copy, forms and a deploy script.

More about websites

One app for macOS and Windows. Free to start; Pro unlocks in the same app.macOS 11 or later · Apple silicon runs it through Rosetta 2.Windows 10 or 11, 64-bit.AgentDock is a desktop app. Open this page on your Mac or PC.

Use the AI you already pay for.

AgentDock brings your existing accounts, signed-in command-line tools and models into one desktop workspace. Every provider is free to use.

  • Available now

    Claude Code

    Anthropic

    Run Claude Code in a project, switch its model and permission mode live, and keep every conversation.

    You bring: Your Claude plan or Anthropic account

  • Available now

    Codex

    OpenAI

    Run Codex through its own SDK, with the model list read from your own install.

    You bring: Your ChatGPT sign-in or OpenAI API key

  • Available now

    Antigravity

    Google

    Google's agent with Gemini models and its own plan mode, beside the others.

    You bring: Your Google account

  • Next release

    NVIDIA NIM

    NVIDIA

    Hosted open models through NVIDIA NIM. Add your key and pick a model; free in AgentDock.

    You bring: Your own NVIDIA API key

  • Next release

    Sarness

    Sarja

    Language, image and video models from your own Sarness server, used directly in the same workspace, free.

    You bring: Your own Sarness install

AgentDock is the desktop front end, not a reseller. You bring compatible access of your own; billing, rate limits and usage are set by each provider. A ChatGPT subscription is not an OpenAI API plan, and buying AgentDock does not include a Claude, ChatGPT, Google or NVIDIA subscription.

Real work, start to finish.

All use cases
  1. 01

    Websites

    Point an agent at a folder, describe the site, review each change before it lands, and keep going until it ships.

    Try asking for: A five-page site for a local business: layout, copy, forms and a deploy script.

    • Claude Code
    • Codex
    • Antigravity
  2. 02

    Full-stack applications

    Uses Pro

    Plan the work, split it across agents, and read what each one changed. With Pro, AIgency plans it once and runs the independent parts side by side.

    Try asking for: A booking app with accounts, a database, payments and tests.

    • AIgency
    • Worktrees
    • Change review
  3. 03

    Desktop software

    Keep the code, the terminal and the agents in one place, and let an agent run builds and read the failures.

    Try asking for: A cross-platform desktop tool with packaging scripts and an auto-update check.

    • Terminal
    • Approvals
  4. 04

    Debugging and code maintenance

    Hand a failing test or a stack trace to an agent, review the diff, revert in one click if it's wrong, and hand the conversation to another agent for a second opinion.

    Try asking for: Find why a nightly job fails, fix it, and upgrade the dependency that caused it.

    • Handoff
    • Change review
    • Terminal
  5. 05

    Product prototyping

    Describe the product, compare how different agents would build it, and keep the version you like.

    Try asking for: Three clickable takes on the same onboarding flow, in a day.

    • Compare
    • Projects

A real, recorded run

Six tasks. Two agents. One working app.

On 10 September 2026, AIgency ran one written brief as a six-task plan across Claude Code and Codex. About ten minutes later there was a command-line todo app that builds, with 58 passing tests as reported by the checking agent.

A small project, chosen so it could be checked, run on a development build. It shows the workflow, not a benchmark.

Read the evidence and its limits
The AgentDock window after the run finished, showing six completed tasks assigned to Claude Code and Codex.

Free connects your AI. Pro puts it to work together.

Pro is for autonomy at scale: AIgency is one capability among many.

  • AIgency

    Preview

    Plan a job once. Run it across your agents.

    Multi-agent planning, execution, scheduling, handoffs and verification. Nothing starts until you approve the plan, and a different agent checks the work.

  • Agent Browser

    Preview

    A browser your agents can read and operate.

    Agents work on real websites through a controlled browser, in view alongside you.

  • Virtual Desktops

    Next release

    A private desktop for autonomous work.

    An isolated desktop an agent can operate without touching yours. Linux first.

  • Automations

    Next release

    Scheduled and recurring agent work.

    Run prompts and workflows on a schedule, with a calendar of what ran.

  • Creative Suite

    Preview

    Production workflows for image, video and voice.

    Advanced media-production workflows, batch automation and orchestration kept with your project. Direct generation through your own Sarness server stays free.

  • Tools & Capabilities

    Next release

    Skills, your own tools and the @ menu.

    Expanded integrations and capabilities, including those from the Arcotheon registry.

  • Advanced orchestration

    Preview

    Concurrency, worktrees and routing.

    Independent tasks in separate git worktrees, agent routing, and recovery when a provider runs out of capacity.

  • Monitoring and control

    Next release

    See what every agent is doing.

    Richer execution visibility, attention management and usage accounting across running work.

See everything in Pro

Free is the whole desktop. Pro is the autopilot.

CapabilityFreePro

Every AI provider

Free is the universal desktop for the AI you already use.

Claude Code
Codex
Google Antigravity and Gemini
NVIDIA NIM, with your own API keyNext release
Sarness models, including image and video generation, on your own serverNext release
Other supported providers, now and laterNext release

Multi-agent orchestration

Your AI working together.

AIgency: plan, run, hand off and verify across agentsPreview
Parallel tasks in isolated git worktreesPreview
Agent routing and capacity recoveryNext release

Autonomy

Agents that operate on their own.

Agent Browser: agents operating real websitesPreview
Virtual Desktops: isolated desktops for autonomous work (Linux)Next release
Automations: scheduled and recurring workflowsNext release

Creative production and tools

Beyond code.

Creative Suite: media-production workflows, batch automation and orchestrationPreview
Expanded tools, capabilities and integrationsNext release
Advanced monitoring, attention and usage accountingNext release

Pro is A$29.99 a month or A$299.99 a year, in Australian dollars. Full comparison and pricing

One app. Start free.

Download AgentDock, sign in with a Sarja Account if you like, and unlock Pro in the same app whenever you want it.

Download AgentDock

One app for macOS and Windows. Free to start; Pro unlocks in the same app.macOS 11 or later · Apple silicon runs it through Rosetta 2.Windows 10 or 11, 64-bit.AgentDock is a desktop app. Open this page on your Mac or PC.