A shared workspace for coding agents

Build with agents.
Together.

Bring your team into the same task. Prompt the agent, add context, and shape the work together, while it happens.

Independent minds. Connected work.
Live multiplayer · Working alphaBuilt by Retia Labs

Meet Plexus · 43 seconds

One agent. More minds.

A look at building together today,
and where shared direction can take us.

Illustrative interfaces. Live multiplayer is in alpha; cloud workspaces and factory coordination are planned.

Explore the workflow
Read the video overview

Plexus brings your team and coding agents into one workspace. The teaser shows three teammates on separate machines contributing to one shared agent task, with code changes visible to the team. It then looks ahead to teams directing multiple tasks and agents at scale. The interfaces are illustrative: live multiplayer prompting works in the alpha, while the complete review and testing workflow is in development and software-factory coordination is a longer-term vision.

One task. Everyone in the loop.

Explore the interactive workflow concept.

Interactive concept
plexusStorefront
One shared task
Storefront / Shared tasks

Checkout recovery

In progress
AlexTask owner
Recover failed checkouts without making a customer pay twice.
CodexWorking
I found the checkout handler. I’m adding a retry path and checking payment safety.src/checkout/retry.ts
Maya joined the task
You’re up to speed.Catch-up

Alex started checkout recovery. Codex has traced the failure path. The retry rules still need a decision.

The conversation, decisions, and source context travel together.

Illustrative workflow. Explore how a task stays shared.

Starting with Codex. Built toward more choice.
Codex First integrationMore agents & models Planned

Shared work starts with shared understanding.

The task is shared.
So is the direction.

Live multiplayer prompting is already working. We’re building the context, handoffs, and review flow around it.

The shared workspace

One task. A shared starting point.

Bring the conversation around the work into the workspace. Teammates can follow the same task instead of reconstructing it from screenshots and separate chats.

Working in the alpha

Everyone can move it forward.

Multiple teammates can prompt the agent directly in the same live task. Add a constraint, ask a question, or help shape the next step together.

In development

Carry understanding forward.

We’re building toward handoffs that preserve the objective, decisions, and outstanding work, so the next teammate has a clear place to begin.

The next milestone

From instructions to reviewed work.

Our next milestone is a reliable shared workflow for inspecting changes, running checks, and reviewing the result before it ships.

A shared rhythm

From a shared idea
to a shared task.

Begin with a coding task on a connected execution device. Bring the objective and the conversation into one workspace.

Coding is our starting point. Codex is our first integration.

Where we’re going

More autonomy.
Still your direction.

We call it controlled autonomy. Agents take on more of the execution. Your team sets the objectives, makes the decisions, and reviews what comes back.

  1. Today · Working alpha

    Build together, live.

    Your team can prompt the same agent in a shared task. We’re strengthening reliability and completing the review and testing loop.

  2. Next · Planned

    Keep the work going.

    Persistent cloud workspaces, where execution can continue after you leave and a teammate can return to review and carry the work forward.

  3. Our longer-term vision

    Direct a software factory.

    Coordinate agents, tasks, and dependencies at a larger scale. People set the objectives, steer the work, and decide whether the result is ready.

Starting with software teams. Growing through the work we do together.

People choose the direction. Agents help take it further.

Controlled autonomy. Shared human judgment.

Closer together. Still in control.

Your workspace.
Your boundaries.

Collaboration should open up possibilities.
You decide what it opens up access to.

Build with us

Share with intention.

We’re designing Plexus around invited teams and explicitly shared projects. Your chosen AI provider receives the inputs it needs to do the work. We’ll explain the available privacy controls before you start using early access.

Your machine. Your agent.

Tasks run on a connected execution host using your supported provider account. Bring teammates into the work while keeping the host and credential ownership clear.

Participation with intention.

We’re shaping a workflow where teams set objectives, review outcomes, and make the calls that need human judgment. Clear ownership and permissions are part of that work.

Join the early-access list

Your next build
is a team sport.

We’re building with small software teams who already use coding agents. Join the early-access list and help shape a workflow your whole team can use.

Early days. Shared ambition.

A little more context.

What is Plexus?

Plexus is a shared workspace for teams and their coding agents. We’re starting with live collaboration: multiple teammates can give instructions to the agent in the same task. Our longer-term vision is controlled autonomy, with teams directing increasingly independent agent work.

What works today, and can I try it?

Live multiplayer prompting works in our alpha. We’re still improving reliability and completing the end-to-end review and testing workflow. Join the list for invite-only early access; we’ll contact teams as places become available. No payment or commitment is required.

Can work continue when I leave?

Today, execution depends on the connected host staying awake and online. Persistent cloud execution and asynchronous review are planned next. Handing responsibility to a teammate does not move a running task to another machine.

Which coding agents will it support?

We’re starting with Codex and your own supported provider account. Broader model and provider choice is part of the direction, with capabilities to be verified for each integration. Provider usage is separate from Plexus.

What does controlled autonomy mean?

Agents can take on more execution while people retain direction: choosing objectives, setting constraints, reviewing outcomes, and intervening when judgment is needed. Coordinating software factories is our longer-term ambition, not a capability of the current alpha.

How are you approaching privacy?

We’re designing controls for who can join a team and which projects are shared. Your chosen AI provider receives the task inputs needed to do the work. We’ll explain data processing, access, and available privacy controls before early access. See Signup privacy below for this website’s signup data.