Hatcher (Bot Builder) — Build Your Own Bot

Hatcher is where you turn "general-purpose AI" into "your dedicated digital employee."

Native Botrun can do a bit of everything, but Hatcher lets you build an AI Bot specialized for a specific business scenario — one that knows your terminology, follows your rules, and only uses the tools you specify.

← Back to Features Overview

Case Study: Ministry of the Interior — Visa Review in 30 Seconds

Scenario: The Department of Civil Affairs reviews large volumes of foreign national study-in-Taiwan applications each year. Each case required 3 hours of manual review.

Hatcher configuration:

• Role: Visa review specialist
• Knowledge base: Eight types of official document draft templates and review form templates
• Tools: PDF analysis (passport recognition), automatic document merging ("paper clip" function)

What the Bot can do:

1. Upload application documents (up to 10-person groups)
2. Automatically recognize names, passport numbers, and dates of birth from passports
3. Automatically verify Taiwan participation ratio (must reach 50%)
4. Generate complete official documents (3 pages) + review forms (1 page)
5. Automatically create a "supervisor review checklist" with page number references

Result: Manual 3 hours → AI 30 seconds to 1 minute

What Is Hatcher?

Hatcher is Botrun's Bot builder. Think of it as an "AI laboratory": you define the role, knowledge, and tools, and Hatcher assembles a custom AI assistant for you.

No coding required. The entire process is just filling out forms, uploading files, and selecting tools.

Native Botrun vs. Hatched Bot

What You Can Configure

Hatcher's configuration is divided into five sections:

1. Basic Settings

2. Prompt Engineering (The Core)

The prompt template is the Bot's "brain instructions" — it defines who the Bot is, how it speaks, and what it can and cannot do.

You can configure the following in your prompt:

3. Bot Mode

When multi-agent mode is selected, the Bot gains "autonomous thinking" — it determines which tools your question requires and combines them automatically.

Model Selection: Basic mode and multi-agent mode each have their own independent model pool, so you can choose the most suitable model for each need.

4. MCP Toolbox (Multi-Agent Mode Only)

Select the tools you want to give your Bot. Currently supports 8 MCP applications:

Multiple tools can be combined. For example: a Bot first uses Perplexity to search for data, then uses Remote PDF to read documents, and finally compiles a report.

5. Knowledge Base

Upload your own documents, and the Bot will answer questions based on this data.

Build Steps (Done in 5 Minutes)

Step 1: Click the "Hatcher" tab in the top navigation bar
  ↓
Step 2: Click "Create BOT"
  ↓
Step 3: Enter a name (e.g., "Ministry of Interior Visa Review")
  ↓
Step 4: Write prompt engineering (define role, rules, tasks)
  ↓
Step 5: Choose BOT mode
  ├── Basic: Pure conversation
  └── Multi-Agent: Can use tools → Select MCP tools
  ↓
Step 6: Upload knowledge base files (optional)
  ↓
Step 7: Test in real-time using the right-side testing panel
  ↓
Done! The system auto-saves every 3 seconds

Save status indicator: The interface shows "Unsaved" → "Saving" → "Saved at XX:XX!"

Share Your Bot

When recipients open the link, they see the Bot's introduction page and can click "Add and Start Conversation" to begin using it.

Advanced Usage

Pairing with High-Taste Examples

Hatcher + High-Taste Examples is the best combination:

1. Create a Bot in Hatcher, defining the role and tools
2. Provide examples using High-Taste Examples (six fields: motivation, benefit, question, answer, source data, evaluation criteria)
3. The Bot automatically learns how to respond from the examples
4. No need to manually refine prompts — once the examples are ready, the prompt engineering is complete

Hatcher FAQ

Next Steps

• ← Back to Features Overview
• Want to learn about native features? → See Native Botrun
• Want to learn about platform capabilities? → See Platform Capabilities