Synthia Agents

Your next coworker has its own computer.

AI coworkers that browse, file, schedule and follow up for you, and ask before anything that matters.

Its own computer

It works on the web the way you do.

Every coworker gets a private browser computer with its own files and a shell. Watch its screen live, and take the mouse and keyboard whenever you want.

Message it like a colleague

Talk to it where you already are.

Write to it in the workspace, from Slack, Microsoft Teams, LINE or WhatsApp, or call it by voice. It replies like a teammate and tells you when it's done.

Approvals

It asks before it acts.

Consequential steps wait in your approvals with a plain summary and a risk level. Passwords and payments are always handed to you.

Routines and responsibilities

Show it once. It takes it from there.

Record a task on its live screen and turn it into a skill or a schedule. Routines run on their own, and responsibilities start when an event arrives from Slack, email, GitHub or a webhook.

Knowledge and memory

It knows your work, and shows its sources.

Give it knowledge shelves for a project or for one coworker. It searches them on every run, remembers what you tell it, and keeps where each memory came from.

Capabilities

Everything a coworker needs to get work done

A computer per coworker

A private browser, files and a shell, with a live screen you can take over.

Approvals

Each request comes with a plain summary and a risk level. Approve or deny in one tap.

Routines

Ask in plain words for something on a schedule, and see every run in one place.

Responsibilities

A lasting goal that wakes up when an event arrives from Slack, email, GitHub or a webhook.

Knowledge

Shelves of files per project or per coworker, searched on every run.

Tasks

A personal task list with due dates and priorities, kept in your workspace.

Teams of coworkers

Several coworkers and people in one thread, passing work along with signed hand-offs.

Voice calls

Call a coworker in its own voice, and share your screen while you talk.

How it works

From one sentence to real work in three steps

  1. Describe the job

    Write one sentence. The builder drafts its name, instructions and suggested skills for you to edit.

  2. Set its boundaries

    Choose its tools, where it may go on the network, and how often it asks you first.

  3. Put it to work

    Message it, give it a routine, and approve what matters. Everything it does is recorded.

Safety and control

You decide how much it does on its own

Governance is part of the product, not a layer on top of it.

Autonomy you set

Ask first, ask only for new kinds of action, or use judgement: chosen per coworker.

Network boundaries

Allow all, an allowlist, or nothing at all, for the whole organization or one group.

Audit trail

Every action a coworker took and every one policy refused, exportable as CSV.

Single sign-on

Require SSO for everyone, with identity provider secrets encrypted at rest.

Private sign-in

You type logins and one-time codes in a private form the coworker never sees.

Signed hand-offs

When one coworker passes work to another, the hand-off is signed and verified.

By default

What it won't do

Each of these is a control in the code, not a promise in a prompt.

Type your passwords

A field or setting named as a password, passcode, two-factor or verification code is handed to you. You sign in through a private form the coworker never sees, and password fields are hidden from everything it reads back.

Pay for something

A pay, checkout, purchase or place-order button, or a card number, CVV or IBAN field, is handed to you. No rule you or your team saves can let the coworker do it.

Reach cloud metadata addresses

Whatever the network policy says, even allow all, the computer refuses link-local and cloud metadata addresses, where a machine's credentials are handed out.

Questions

What people ask first

What is an AI coworker?

A named assistant with its own instructions, tools and computer. You message it like a colleague, and it carries a job through to the end.

Will it act without asking me?

Only as far as you allow. You choose its autonomy level, and passwords and payments are always handed to you.

Where can I talk to it?

In the workspace on the web or the mobile app, from Slack, Microsoft Teams, LINE or WhatsApp once they are linked, and by voice call.

Can I see what it did?

Yes. Watch its screen live while it works, and read the audit trail afterwards: every action, and every one that policy refused.

Which AI models does it use?

The providers your administrator connects. The model providers guide lists them and how to switch.

Model providers guide →

Put your first coworker to work

Sign in, open a channel, and ask it to do something real on the web.

Open the workspace→