> ## Documentation Index
> Fetch the complete documentation index at: https://docs.joystream.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# How JoyStream thinks

> The belief, the category, and the primary object — the chain of reasoning the whole product rests on.

Every organization is a collection of capabilities: the recurring work it must do well. Onboarding employees. Qualifying inbound leads. Closing the books. Investigating incidents. Reviewing contracts.

Historically these lived across people, processes, documents, SaaS applications, institutional knowledge, and informal coordination. Increasingly, they become software.

That does not mean organizations become autonomous, or that one general-purpose agent runs the company. It means recurring organizational work is increasingly represented through explicit specifications, intelligent agents, executable plans, operational memory, evidence, and human governance.

## The chain

```mermaid theme={null}
flowchart TD
    A["<b>Belief</b><br/>Every organizational capability is becoming software,<br/>built and operated by human and AI teams"]
    B["<b>Category</b><br/>Agent Engineering"]
    C["<b>Promise</b><br/>Agents should earn their way into production"]
    D["<b>Mission</b><br/>Build the operating system through which human and AI teams<br/>turn organizational work into Capabilities they can own"]
    E["<b>Product</b><br/>JoyStream"]
    F["<b>Primary object</b><br/>Capability"]
    A --> B --> C --> D --> E --> F
```

**The vision.** A world where human and AI teams continuously engineer, execute, and improve every organizational capability.

## What changes, and what doesn't

| Today a capability is   | Tomorrow it is                            |
| ----------------------- | ----------------------------------------- |
| People                  | A human team                              |
| Process                 | AI agents                                 |
| Documents               | A capability spec                         |
| SaaS tools              | Runtime recipes                           |
| Meetings                | Deployments                               |
| Institutional knowledge | Operational memory, evidence, and metrics |

The future organization is not operated by one omniscient agent. It is operated through many specialized capabilities, owned and improved by human and AI teams.

People remain accountable for defining intent, setting boundaries, approving access, reviewing evidence, handling exceptions, resolving ambiguity, governing production, and improving the capability over time. Agents perform work inside that system.

## Why the capability, not the agent

An agent trapped on one person's laptop is an organizational liability. When that person leaves, the agent and its context leave with them.

But organizational capabilities were never one person's to begin with. Lead qualification, onboarding, customer escalation, finance — these are owned by teams. They have owners, operators, contributors, reviewers, approvers, stakeholders, and beneficiaries.

```mermaid theme={null}
flowchart LR
    A["<b>Personal agent</b><br/>One person, one prompt<br/>One conversation, one laptop<br/>Fragile ownership"]
    --> B["<b>Team capability</b><br/>Shared agent and plan<br/>Shared runs and tickets<br/>Shared evidence"]
    --> C["<b>Organizational capability</b><br/>Durable, versioned, governed<br/>Deployed, operated, measured<br/>Continuously improved"]
```

The capability is the asset. The agent is one implementation of it.

## The three surfaces

JoyStream has three product experiences. They are not independent products — they are three views of one continuous lifecycle.

<Columns cols={3}>
  <Card title="Joyvis" icon="message-circle">
    Engineer the capability. A conversation that produces a structured, versioned, reviewable, executable spec.
  </Card>

  <Card title="Workspace" icon="users">
    Own the capability together. Teams, versions, reviews, approvals, permissions, and sharing.
  </Card>

  <Card title="Operations" icon="activity">
    Run and improve the capability. Runs, work, tickets, evidence, metrics, and intervention.
  </Card>
</Columns>

Joyvis is the front door. Workspace establishes durable organizational ownership. Operations is where the capability performs real work, accumulates evidence, and becomes more valuable.

## The product test

Every product decision should strengthen at least one of these. Does it help a team **discover** the right organizational job, **specify** a capability with clarity, **engineer** an agent with rigor, **bound** its intelligent behavior, **prove** its outcome through evidence, **promote** it safely, **operate** it in the real world, **see** its work, **collaborate** around shared ownership, **improve** it through execution, or **share** it without losing context?

## Next

<Columns cols={2}>
  <Card title="Agent Engineering" href="/concepts/agent-engineering" icon="hard-hat">
    The category, and why generation isn't the hard part.
  </Card>

  <Card title="Capabilities" href="/concepts/capabilities" icon="box">
    The primary object and everything that hangs off it.
  </Card>
</Columns>