Creating Agents
AI Prompt Builder
The fastest way to create an agent. Describe what you want in plain English and let AI do the rest. No coding required.
How It Works
The AI Prompt Builder uses Google Gemini to generate a complete agent configuration from your natural language description. Go to Create → AI Generator to get started.
Step 1: Describe Your Agent
Write a description of what you want your agent to do. Be as specific as possible. Good descriptions include:
- The agent's area of expertise
- Specific tasks it should handle
- Who the target audience is
- Any particular approach or methodology
Example Prompts
Here are some effective prompts to inspire you:
Data Analysis Agent
"Create an agent that analyzes CSV data, finds patterns and trends, generates statistical reports with charts, and provides actionable business insights."
API Integration Builder
"Build an agent that helps developers test REST APIs, validate responses, generate documentation, and troubleshoot integration issues."
Contract Analyzer
"Design an agent that reads legal contracts, identifies key clauses, highlights risks, compares terms against industry standards, and generates plain-language summaries."
SEO Auditor
"Create an agent that audits websites for SEO issues, analyzes page content, checks technical SEO factors, and provides prioritized improvement recommendations."
Step 2: Refine the Configuration
After AI generates the initial configuration, you can refine:
- Agent Name — Edit the auto-generated name (aim for 2-4 catchy words)
- Description — Tweak the value proposition shown to users
- System Prompt — Review and adjust the generated instructions. This is the most impactful thing you can edit.
Step 3: Configure Tools
Select from 17 available tools across three tiers. The AI may auto-select tools based on your description, but you can add or remove them:
| Tier | Tools |
|---|---|
| Free | web_search, markdown_table |
| Standard | web_scrape, csv_parse, data_query, statistics, generate_chart, data_transform, regex, document_parse, template_render, text_diff, json_extract |
| Premium | api_call, code_execute, email, calendar |
For full tool documentation, see the Tools Reference.
Step 4: Set Pricing & Deploy
Choose your pricing model and visibility, then deploy:
- Pricing — Free or Per Task ($0.05 – $9.99). Premium tools justify higher prices.
- Visibility — Public (marketplace) or Private (direct link only)
Click Deploy Agent and your agent goes live. See the Publishing guide for what happens next.
Writing Effective System Prompts
The system prompt is the single most important factor in agent quality. A great system prompt includes:
- Role definition — "You are [name], an expert in..."
- Specializations — Bullet list of specific areas of expertise
- Guidelines — How to handle requests, tone, level of detail
- Tool instructions — When and how to use available tools
- Limitations — What not to do, when to defer to humans
Example System Prompt
You are Data Analyst Pro, an expert AI data analysis assistant.
You specialize in:
- Spreadsheet and CSV data analysis
- Statistical analysis with clear explanations
- Data visualization recommendations
- SQL query writing and optimization
- Business intelligence reporting
- Data cleaning and preparation
Guidelines:
- Ask clarifying questions about the data and goals
- Use csv_parse to read data, then statistics for analysis
- Always generate_chart when presenting numerical findings
- Provide clear, actionable insights — not just numbers
- Explain statistical concepts in plain language
- Identify trends, anomalies, and correlations
- Format results with markdown_table for readability
- Recommend concrete next steps based on findings
Important: When analyzing data, always start by understanding
the structure (columns, types, row count) before diving into
analysis. Present findings in order of business impact.Next Steps
- Templates — Start from a production-ready template instead
- Tools Reference — Detailed documentation for all 17 tools
- Publishing — Deploy and list your agent on the marketplace