NYPTIDMeridian

Meridian ADE

The agent workspace.

Give an agent a task and a repository. It plans, edits, runs the code, checks the result in a real browser, and hands you the evidence. Run several at once and watch all of them from one dashboard.

From $80/month · Every frontier model · No API keys

Get ADE Try the live demo

Durable runs

Kill it mid-run. It picks up where it stopped.

Every step of an agent run is written to a journal before it happens. If the host restarts, the run comes back paused with its history intact and resumes from the exact step, without repeating a write it already made.

It checks its own work

Agents look at what they built.

An agent can load the page it just changed in a headless browser and read the screenshot, or call the endpoint and read the response, before it tells you it works.

Review gate

Nothing is done until you say so.

A run ends by proposing a result with evidence. It waits at "Ready for review" until you mark it verified or send it back with a note.

Agents dashboard

Every agent, one glance.

See which agents are working, which finished, and which are waiting on you, with how long each has been at it. Click one to bring it into a pane.

Workspaces

Sessions live under their project.

The sidebar groups sessions by the project they belong to, with live counts of what's running and what needs you. Collapse what you're not working on.

Up to four panes

Watch four agents side by side.

Each pane has its own session, model and timer. Switch between one, two, three or four panes without losing anything.

Loop or chat

A full agent run, or a quick answer.

Run a session as a durable agent loop, or as a fast chat when you just want to ask something. Switch mid-conversation.

Every frontier model

Pick the model per session.

Claude, GPT, Grok, Gemini, Kimi, DeepSeek, Qwen and more on one subscription. Change models without changing tools.

Memory & skills

It learns your project. You approve what it keeps.

Agents propose facts and reusable procedures as they work. Nothing enters their context until you approve it.

Routines & connections

Work on a schedule, tools on a leash.

Put an objective on a schedule and find the result on your board. Connect GitHub and any MCP server you run; every outside call asks first.

Where it runs

Your agents keep working when your PC sleeps.

Agent runs execute on Meridian's cloud host. Open the workspace in your browser, close the lid, and come back to finished work waiting for review.

Open ADE
Agent runsOn Meridian's cloud host, journaled step by step
Your codeConnect GitHub and agents work on the repositories you choose
ApprovalsFile writes and every external tool call can wait for your OK
ModelsFrontier models from Anthropic, OpenAI, xAI, Google, Moonshot, DeepSeek, Qwen and more
API keysNone. Usage comes out of your plan's included credits

ADE plans

Start with one agent. Scale to many.

Every plan includes cloud credits for model usage. Top up any time.

Start here ADE $80/ month

Billed monthly.

The full agent workspace for one developer.

  • 8,000 Cloud credits / month
  • $50 included inference
  • 3 agents running at once
  • 5 workspaces
Choose ADE
ADE Plus $160/ month

Billed monthly.

For keeping several agents busy on the same repo.

  • 20,000 Cloud credits / month
  • $120 included inference
  • 8 agents running at once
  • Unlimited workspaces
Choose ADE Plus
ADE Max $280/ month

Billed monthly.

For days with many agents running at once.

  • 45,000 Cloud credits / month
  • $220 included inference
  • Unlimited agents at once
  • Unlimited workspaces · priority routing
Choose ADE Max

Want voice too? The Meridian bundle includes Echo and ADE from $100/month.

Put your first agent to work today.

Pick a plan, open a workspace, and hand it the task you've been putting off.

Get started Open ADE