Skip to main content
Distillation is the core of creating on Profy — through a structured AI conversation, the system deeply interviews you and extracts your professional knowledge into a complete AI Agent.
Distillation isn’t filling out a form. It’s like an in-depth expert interview — you just express your thinking naturally, and the AI structures it for you.

​
What Is Distillation?

Imagine you’re teaching an extremely intelligent apprentice. This apprentice isn’t satisfied with “what to do” instructions — it wants to understand “why you do it this way,” “when not to do it this way,” and “how you handle edge cases.” Distillation is this process: the AI acts as an interviewer, using a carefully designed question framework to systematically extract your:
  • Mental Models: How you understand and organize knowledge in this domain
  • Decision Heuristics: Your judgment logic when facing complex situations
  • Expression Style: Your unique communication style and language habits
  • Anti-pattern Awareness: What you know to be common mistakes, what paths to avoid
  • Honesty Boundaries: Where you’re clear about the limits of your capabilities

​
Five Layers of Cognitive Extraction

The distillation AI explores conversation around these five layers:
1

Mental Models

How do you understand this domain? What are the core concepts and their relationships? What does the “knowledge map” in your head look like?
2

Decision Heuristics

What’s your thinking path when facing typical problems? What signals drive a particular judgment? What “if…then…” rules do you follow?
3

Expression DNA

How do you explain complex concepts to people at different levels? What’s your tone, rhythm, and go-to metaphors? What makes your “voice” unique?
4

Anti-patterns

What are the most common mistakes in this domain? What pitfalls do beginners most often fall into? What are the worst practices you’ve seen?
5

Honesty Boundaries

What don’t you know? What questions exceed your capabilities? When would you advise someone to seek help elsewhere?

​
Starting Distillation

1

Enter Studio

In your creator workspace, click “Create New Expert” or “Start Distillation.”
2

Choose a Domain

Briefly describe the professional direction of the Expert you want to create. This helps the AI adjust its questioning strategy.
3

Upload Reference Materials (Optional)

You can upload documents, ZIP files, project configuration files, etc. as background reference. The AI will combine these materials to ask more targeted questions.
4

Begin the Conversation

The AI will start interviewing you. Just answer naturally, as if chatting with a curious professional peer. The process typically takes 15–30 minutes.
5

Scenario Calibration

The AI will test the extracted knowledge against a few typical scenarios. If the responses don’t match your expectations, you can correct them directly.
6

Generate Expert

After distillation is complete, the system automatically generates a full Expert including persona, skills, and configuration. You can continue refining it in Studio.

​
Cross-Platform Import

If you’ve already created AI assistants or agents on other platforms (such as OpenClaw, Claude Code, Cursor, etc.), the distillation AI can automatically recognize and extract relevant configurations. Simply mention or upload related files during the conversation, and the AI will help migrate your existing settings to the Profy Expert format.

​
Distillation Tips

Before distillation, prepare 3–5 real cases you’ve handled. Specific cases are more effective than abstract descriptions — the AI can extract decision patterns you haven’t explicitly stated.
Don’t just say “I’d recommend X.” Instead, say “I’d first look at A and B; if A meets condition C, I’d recommend X; otherwise I’d consider Y.” Expose your reasoning chain.
A good Expert must know its boundaries. Clearly tell the AI about areas you’re uncertain about — this makes your Expert more credible.
Tell the AI “a common mistake is…” or “if someone tells you X, that’s usually wrong.” Anti-pattern knowledge is what separates beginners from experts.
The distillation AI will structure things for you. Just express yourself as naturally as you would talking to a colleague. Don’t worry about perfect wording.

​
After Distillation

Once distillation is complete, your Expert already has basic capabilities. From here you can:
  • Test Conversations: Chat with your Expert in Studio to verify its responses meet your expectations
  • Adjust Configuration: Fine-tune the persona, enable tools and plugins
  • Install Skills: Install additional skills from the marketplace to extend capabilities
  • Set Pricing: Choose the right pricing model for you
  • Publish: Submit for review and go live
Distillation isn’t a one-time event. You can return to Studio anytime to supplement and refine your Expert through ongoing conversation. After publishing, the Darwin evolution system will also automatically learn from real user interactions.

​
Next Steps

Configure Your Expert

Tools, skills, and plugins — full configuration

Set Pricing

Choose the right pricing model