# What Connect is

> **What you will learn**
> What kind of product Connect is, how it differs from HyperTeams, and how the
> two link up.

## In one line

> **An AI workspace that a team shares.**

You open it in a browser, direct work through conversation, and manage the tools
and the workers that handle it **at the team level.**

## How it differs from HyperTeams

The two are easy to confuse, so:

| | HyperTeams | HyperTeams Connect |
|---|---|---|
| Where it runs | your machine | **the cloud** |
| Who uses it | you | **the whole team** |
| What it works on | your actual files | a workspace and connected tools |
| Data | stored on your machine | stored in the cloud |
| Sign-up | not needed | required |
| Strength | works directly on your files | **sharing, collaboration, permissions** |

**In one line**: working alone on your own files is HyperTeams; a team working
together is Connect.

## The two connect

They are not competitors.

```mermaid
graph TD
  T["Team members"] --> C["Connect<br/>the cloud"]
  C --> S["System registration"]
  S --> H["HyperTeams<br/>your machine"]
  H --> F["Real project files"]
  C --> M["MCP tools"]
  C --> A["Agents"]
```

[Register your machine as a system](/guide/cn-systems) in Connect and you can
direct work on that machine from a team workspace. **A path opens from the
team's request to the real files on your machine.**

## What Connect solves

Moving from personal AI use to a team creates new problems.

| Alone | As a team |
|---|---|
| You know what works | **nobody else does** |
| Your conversation history | invisible to others. Everyone asks the same question |
| Tools you attached | teammates have to attach them again |
| Your cost | nobody knows who spends what |
| Being careful is enough | **you need permission management** |

Connect addresses these five.

| Problem | Connect's answer |
|---|---|
| What works is not shared | **agents** — write the instruction once and publish it |
| History is scattered | **workspaces** — conversations in one place |
| Everyone attaches their own tools | **MCP connections** — attach once, the whole team has it |
| Costs are invisible | **the usage screen** — who spent what on what |
| Permission management | **roles, groups, departments** |

## What it is made of

Learn four things and most other screens become guessable.

```mermaid
graph TD
  W["Workspace<br/>the team's boundary"] --> T["Thread<br/>one job = one conversation"]
  W --> A["Agent<br/>the thing that does the work"]
  W --> S["Settings<br/>tools · members · usage"]
  T --> A
  A --> TL["Tools<br/>MCP · built-in · systems"]
  TL --> EX["Outside<br/>APIs · your machine · files"]
```

Each gets its own chapter next.

## Common questions

### "Do I need both?"

No. **Use only what you need.**

- Solo code work → HyperTeams only
- A team doing work with AI → Connect only
- A team reaching into individual machines → both

### "How is this different from the AI service we already use?"

The biggest difference is **tool connections and permission management.**
Consumer services converse well, but connecting to internal systems and
splitting permissions by team is a different problem.

### "I'm worried about data being stored in the cloud"

There is an air-gapped (on-prem) install. In that case data does not leave the
organisation. The specific terms need a consultation.

---

## Check yourself

**1. What is the biggest difference between HyperTeams and Connect?**

<details>
<summary>Answer</summary>

**Where it runs and who uses it.** HyperTeams runs on your machine and you work
on your own files alone; Connect runs in the cloud and a team uses it together.
</details>

**2. What five problems appear when moving from solo to team use?**

<details>
<summary>Answer</summary>

**What works is not shared, history is scattered, everyone attaches their own
tools, costs are invisible, and permission management is needed** — Connect's
agents, workspaces, MCP connections, usage, and roles each answer one.
</details>

**3. How do the two products connect?**

<details>
<summary>Answer</summary>

**Register your machine as a system** in Connect and you can direct work on that
machine from a team workspace.
</details>

---

Starting with the most basic unit → [Workspaces](/guide/cn-workspace)