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

# Expert

> Every Expert is a complete autonomous agent instance

Expert is the core product unit of the Profy platform. Each Expert is a **complete autonomous AI agent instance** — it's not just a chat window, but an intelligent entity with its own persona, skills, memory, tools, and independent workspace.

## Expert vs Regular Chatbots

| Dimension     | Regular Chatbot           | Profy Expert                                                                                 |
| ------------- | ------------------------- | -------------------------------------------------------------------------------------------- |
| Persona       | Generic, undifferentiated | Created through cognitive distillation, with a unique mental model and expression style      |
| Memory        | Current conversation only | Persistent cross-session memory, remembers your preferences and past decisions               |
| Tools         | Limited or none           | 20+ built-in plugins for file processing, code execution, website building, media generation |
| Workspace     | None                      | Independent cloud sandbox for writing code, previewing websites, and managing files          |
| Evolution     | Static and unchanging     | Darwin self-evolution system, continuously improving skills and behavior                     |
| Collaboration | Single                    | Supports multi-Expert delegation and collaboration                                           |

## Components of an Expert

An Expert consists of the following parts:

<CardGroup cols={2}>
  <Card title="Persona" icon="mask">
    The Expert's core identity — who it is, how it thinks, its areas of expertise, and its expression style. Set through cognitive distillation or manual configuration
  </Card>

  <Card title="Skills" icon="wand-magic-sparkles">
    Domain-specific capability instructions. A writing Expert might have skills like "long-form writing" or "copywriting optimization"
  </Card>

  <Card title="Memory" icon="brain">
    Persistent memory of user preferences and historical interactions. The Expert remembers your past choices and automatically applies them next time
  </Card>

  <Card title="Tools & Plugins" icon="toolbox">
    The set of tools available to the Expert — search, file processing, code execution, image generation, and more
  </Card>

  <Card title="Sandbox Workspace" icon="cube">
    An independent cloud environment allocated for each conversation, used for code execution, file management, and website preview
  </Card>

  <Card title="Guard Policy" icon="shield">
    Defines what the Expert should not do — intellectual property protection and behavioral boundaries
  </Card>
</CardGroup>

## Expert Lifecycle

From creation to user interaction, an Expert goes through the following stages:

<Steps>
  <Step title="Creation">
    Creators build an Expert through two methods:

    * **Cognitive Distillation**: Upload professional materials (documents, conversation logs, etc.), and AI automatically extracts the cognitive structure
    * **Manual Configuration**: Manually set up persona, skills, and tools in the Studio

    At this point, the Expert is in **draft** status, visible only to the creator.
  </Step>

  <Step title="Review for Publication">
    After the creator submits a publication request, the Expert enters **review** status. The platform reviews content quality and compliance.

    * First publication: After approval, the Expert is listed on the marketplace
    * Version updates: The old version stays online while the new version is reviewed independently — users are not affected during updates
  </Step>

  <Step title="Live on Marketplace">
    After approval, the Expert enters **published** status and becomes visible to all users on the marketplace.

    Creators can save draft changes at any time (without affecting the live version) and adjust pricing strategies.
  </Step>

  <Step title="Continuous Evolution">
    Once listed, the Expert continuously improves through the Darwin evolution system:

    * Automatically extracts valuable memories from conversations
    * Periodically evaluates skill quality and produces improvement suggestions
    * After creator review and acceptance, skill improvements take effect immediately
  </Step>
</Steps>

## Expert Pricing Models

Creators can choose different pricing strategies for their Expert:

| Model                 | Description                                                    | Use Case                         |
| --------------------- | -------------------------------------------------------------- | -------------------------------- |
| **Free**              | Users only consume their own credits (AI call costs)           | Open-source tool-type Experts    |
| **Trial + Paid**      | Offers a certain number of free trials, then requires purchase | Let users try before they decide |
| **One-time Purchase** | Pay a one-time fee for permanent access                        | High-value professional Experts  |
| **Usage-based**       | Charged per AI call usage                                      | High-frequency everyday tools    |

<Note>
  Specific prices are set by creators. Refer to the Expert detail page in the product for real-time pricing.
</Note>

## Key Expert Attributes

### Identifier

Each Expert has a globally unique identifier, used for API calls and URL construction. The identifier is set at creation time and cannot be changed afterward.

### Category

Experts are categorized by domain (e.g., writing, programming, design), and users can filter by category in the marketplace.

### Version

Each published update creates a new version. Users always use the latest approved version, with historical versions retained for reference.

### Featured

Platform editors can mark high-quality Experts as "featured" for prominent display on the marketplace homepage.

## Expert App: Standalone Product Mode

An Expert can be further configured as a standalone App — with its own subdomain (`xxx.profy.cn`), independent user interface, and admin panel. This elevates an Expert from a marketplace agent to an independent AI product.

<Tip>
  Expert App pricing, admin panel, and user system differ from regular Experts. When your Expert needs its own brand, you can enable App mode in the Studio.
</Tip>

## Next Steps

<CardGroup cols={2}>
  <Card title="Sessions & Workspace" icon="comments" href="/en/documentation/concepts/sessions-and-workspace">
    Learn about session and workspace mechanics when chatting with an Expert
  </Card>

  <Card title="Skills" icon="wand-magic-sparkles" href="/en/documentation/concepts/skills">
    Learn how to give Experts domain-specific capabilities through skills
  </Card>

  <Card title="Plugins" icon="plug" href="/en/documentation/concepts/plugins">
    Explore the 20+ built-in plugins available to Experts
  </Card>

  <Card title="Memory" icon="brain" href="/en/documentation/concepts/memory">
    Learn how Experts remember user preferences and historical context
  </Card>
</CardGroup>