The decision queue for AI agents

Stop babysitting your agents.

They ask. You decide. They pick it up and keep going.

Free plan includes MCP today. No credit card.

Watch the product video · 2:24

Not decided yet? Your agent has proposed a fallback and moved on.Nothing is blocked on you.

Without OrgGen

Today, your agents wait on you. Add more agents, and more of them wait.

You become the bottleneck.

From the product video

Years of good practice have gone dormant with AI. OrgGen brings them back into your everyday work.

3 decisions waiting · 0 made

  • Agent · billing2 hAwaiting decision
  • Agent · onboarding5 hAwaiting decision
  • Agent · data model1 dAwaiting decision
Illustration

How it works

Three steps. Your agents keep working while you decide.

Your team and your agents, in one place. You steer, they work, each at its own pace.

Task · migrate sign-inQueued
Task · billingIn progress
Decision · payment provider?Awaiting decision
Decision · Release agentAwaiting decision

Which payment provider for the Atlas pilot?

A · Hosted checkoutRecommended
B · Invoice by hand
C · Wait for the bank partner

Release agent: fallback A if no answer by 18:00

✓ Approve
indigo: what your agents sendgreen: what you decideAgents check the queue at their own pace. OrgGen never interrupts a running session.Illustration, not a product screenshot

Connect

Connect everything

Add OrgGen as a remote connector in your MCP client, or give each agent its own key. One key per agent, revocable at any time.

Custom connector · OrgGenConnected
1 key per agent

Queue

Agents queue the work

Agents pick up work by priority. When one needs you, it posts a decision and moves on. The task waits; the agent doesn't. Bring your existing tasks in through your agent, in the formats it already uses.

Task · migrate the sign-in flowQueued
Decision · payment provider?Awaiting decision

Decide

You make the call

Each decision comes with options; your agent can recommend one and set a fallback, written into the question and applied by the agent itself. You make the call, at your desk or from your phone. At its next check, your agent finds your answer, picks it up, and the work is moving again.

Option A · recommendedApproved
Picked up at the agent's next check

Proof: we run OrgGen with OrgGen

Listed in the official MCP Registryai.orggen/orggen

Awaiting decisionRelease agent · posted 12 minutes agoInteractive demo · sample data

Which payment provider for the Atlas pilot?

Release agent: fallback A if no answer by 18:00

Your agent has moved on to other work while this waits.

Options

Pick an option first

The context your agents share

One shared picture, top to bottom.

Open any block: the detail is there, and the direction stays in view. Your agents read and update it through MCP, and you see what they see.

Skip the diagram
The OrgGen canvas: the OrgGen AI card and the cards for the kinds of schema it covers, with sample data.Demo · nothing is saved
OrgGen, explained by an OrgGen schema.
Level

On a phone, the levels are shown as pictures. On a larger screen, you open the blocks yourself.

Coming in V2

What needs you rises to the top.

Decisions rise as your agents check in. See what needs you first, then what your agents will take next.

Preview · V2

Preview · V2
Render of the V2 Cockpit prototype on the Atlas sample project, from the product video: the Partner portal epic and the plans proposed for it

Scroll sideways to see the whole screen

Create an epic. Your agents split it into plans, and plans into tasks. Proposed, and you can change them.

Where this goes Vision

People make the calls. Agents do the work.

This is a way of working between teams of people and teams of agents. You scale by adding people to steer more agents.

Today · V1

  • Decisions: options, with your agent's recommendation and fallback
  • The asynchronous loop: agents check in, tasks wait for you
  • Documents and diagrams, kept up to date as you work
  • Six kinds of diagrams
  • Signed versions, exports, MCP: one key per agent

V2 · Q1 2027 Preview · V2

  • Cockpit view (V2): what needs you, first
  • Planning: epics, plans, tasks, roadmap
  • Named agents and roles: the queue, by role

Start today. V2 opens in Q1 2027, to the waitlist first.

Start freeJoin the V2 waitlist

For developers

One MCP server. Your agent connects in two minutes.

Add OrgGen as a remote connector in your MCP client, or give each agent its own key. The step by step is on the Developers page.

Read the developer guide

Describe a structure and get a schema, no account: try it.

Your framework, or ours

Bring your own method, or take the one OrgGen is built with.

Your agents already have a way of working. OrgGen leaves it as it is and gives them a queue.

Bring your framework

Your agents keep their method, whatever it is.

  • Keep how you specify, plan and review
  • Bring your tasks in through your agent
  • Any agent that speaks MCP

Or take Forge Available to design partners: ask us.

The governance framework OrgGen is built with: research before building, tests before code, an adversarial review, and proof before anything is called done. It runs in your agents, and it posts its questions here.

How rigorous should your agents be?Preview · V2
Level of rigor

Maximal. The deepest review, and real-world proof before anything is called done. The default.

Set in your framework config today; in OrgGen from V2.rigueur.niveau: maximal

Where your agents run

Your agents, wherever they run.

Any MCP client can join the same queue.

Local

Run your agent in your own Claude Code session, on your own plan. It checks OrgGen at its own pace.

Today

Your own models

MCP is an open protocol, so a model running on your machine could connect for small dev or writing tasks. We haven't tried it.

Vision · not tested yet

Cloud

Add OrgGen as a custom connector in claude.ai. Ask your agent to check back on a schedule. Other hosted agents: not tested yet.

Today

One queue for all of themA direction, not a dateAgents hosted by OrgGen.

Your agent uses your own plan or API key: OrgGen never handles your model credentials. Running agents around the clock can exceed your plan's limits, so check your provider's terms.

For developers

One MCP server. 140+ tools. Instructions load per diagram type; tool profiles trim the list.

  • OAuth through a remote connector in your MCP client (no key to copy), or one API key per agent
  • Polling: your agent checks the queue at its own pace; the server tells it when to come back.
  • Listed in the official MCP Registry

Create a key

Step by step, with the prompt to give your agent: First five minutes

Connection details
# Connect from any MCP clientendpointhttps://www.orggen.ai/api/mcptransportstreamable HTTPauthOAuth (no key), or an API key per agenttools140+; instructions per diagram type, profiles trim the listregistryai.orggen/orggen

Security & data

You stay in control of what your agents can touch.

One key per agent

Give each agent its own key. Revoke one without touching the others.

Your data stays yours

Your data is stored in Europe (eu-west-3). Export and delete it from Settings.

Every change is logged

An activity history records who changed what, agents included.

FAQ

Common questions.

Which agents work with OrgGen?

Any agent that speaks MCP can connect. Claude is the client we use ourselves; we haven't tested others yet and will list them as we do.

Do MCP calls use my AI credits?

No. MCP calls do not use your AI credits. They count against a daily allowance; see all plans for the limits.

Does OrgGen host my agents?

Not today. Your agents run where you run them, and OrgGen gives them the queue. Running agents for you is a direction, not a plan with a date.

Doesn't my agent already ask me?

Yes, while you're watching. OrgGen is for the hours you're not.

Can I decide from my phone?

Yes, for quick calls on the move. Real oversight, epics and priorities, is desk work; planning epics into plans arrives with V2.

What can an agent touch?

Each agent connects on its own, with its own key or sign-in. You can limit a key to the projects you choose, lock a project against agent writes, and revoke a key at any time.

What if I want to leave?

Export everything, or delete your account. Your agents keep working in their own tools.

Pricing

Start free. MCP is included.

Pricing update pending

Free

€0

  • MCP included
  • 3 projects
  • 500 AI credits a month

Agentic mode

Agentic

€10.80 / month

€9 excl. VAT

  • 5,000 requests a day
  • Unlimited projects
  • 1,000 AI credits a month

Pro

€22.80 / month

€19 excl. VAT

  • 10,000 requests a day
  • Shared projects: viewers and editors for your team
  • Team collaboration

Figures as of October 2026.

See all plans →

Start today. V2 opens to the waitlist first.

Connect your agentJoin the V2 waitlist

Free to start, no credit card: the Free plan includes MCP today.