For AI agents: a documentation index is available at the root level at /llms.txt and /llms-full.txt. Append /llms.txt to any URL for a page-level index, or .md for the markdown version of any page.
Templates give you a head start when building Agents and Flows. Instead of starting from a blank canvas, you can pick a template that is already configured with steps, tools, and settings, then customize it to fit your needs.
Flow Templates
Flow Templates are pre-built workflows organized by category, so you can quickly find one that matches what you want to build.
How to create a Flow from a template
Click Flows in the sidebar.
Click Create Flow.
In the Create New Flow panel, click Use an example.
Browse templates by category or scroll through the list. Each template card shows the title, a short description, and the number of steps included.
Click a template to select it. The Flow editor opens with the name, description, steps, and variables already loaded.
Templates are a good way to learn how Flows work. Open one, review each step, and see how they connect before building your own.
Available Flow Templates
Support
AI Support Agent — Automated support ticket handling with intelligent routing and response generation.
AI Support Auto-Responder with Knowledge Base — Combines knowledge base search with AI-generated responses for faster support.
Content
Content Creation Pipeline — Fetch content from a URL, analyze it, and generate new content with AI.
Analytics
Data Processing Pipeline — Pull data from an API, transform it, analyze with AI, and store the results.
Transformation
Convert HTML to Markdown — Fetch a web page and convert its content to clean Markdown.
Data Extraction
Contact Information Extraction — Extract and structure contact details from web pages.
Data Processing
Data Encoding and Validation — Retrieve Records, transform data, and validate the output with AI.
Research
Autonomous Research Pipeline — A multi-step pipeline that researches a topic, synthesizes sources, and drafts long-form content.
New templates are added regularly, so check back if you do not see one that fits your use case.
Customize after selection
Once a template is loaded in the editor, you have full control over how it works.
Rename the Flow and update its description
Add, remove, or reorder steps
Swap models, edit prompts, or change step configurations
Adjust variables and metadata
Think of templates as a starting point, then make them your own.
Agent Templates
Agent Templates are organized by type. Each type includes different default settings, tools, and behaviors for common use cases.
How to create an Agent from a template
Click Agents in the sidebar.
Click New Agent.
Choose an Agent type: AI Assistant, Autonomous Agent, Research Agent, A2A Agent (External), or Start with an empty Agent.
Fill in the configuration for your chosen type, then click the create button.
You can also create a new Agent directly from a Product by clicking Add Capability → New Agent.
AI Assistant
An AI Assistant is a conversational Agent with a predefined persona. It works well for chat-based use cases like customer support or Q&A.
Settings:
Name (required) — Give your Agent a descriptive name, such as Customer Support Assistant.
Model (required) — Choose the AI model that powers conversations.
Description — Briefly describe what the Agent does.
Persona template — Pick a preset to get started quickly.
Click Create AI Assistant to finish.
Autonomous Agent
An Autonomous Agent uses multi-step reasoning and iterative planning to work through complex tasks on its own.
Settings:
Name (required) — Agent name.
Model (required) — AI model.
Description — Describe the Agent’s purpose.
Enable Agent Loop — Turn this on to let the Agent plan, act, and reflect across multiple iterations.
Max Iterations — Set a limit on reasoning cycles from 2 to 25 to keep the Agent focused. This appears only when Agent Loop is enabled.
Click Create Autonomous Agent to finish.
Agent Loop works best for tasks where the Agent needs to evaluate its own output and refine it. Start with a lower iteration count, then increase it as you tune your prompt.
Research Agent
A Research Agent comes with web research tools and is ready to search the web and extract content.
Settings:
Name (required) — Agent name.
Model (required) — AI model.
Description — Describe the Agent’s purpose.
Exa — Web search tool for finding relevant sources. Included by default.
Firecrawl — Web scraping tool for extracting page content. Included by default.
Research Cycles — Maximum number of research iterations from 1 to 10.
Click Create Research Agent to finish.
Research Agents work best with both Exa and Firecrawl enabled. Exa finds the sources, and Firecrawl extracts the content.
A2A Agent (External)
Use an A2A Agent to connect an Agent built with another framework through the Agent-to-Agent protocol.
Agent Card URL — Provide this to auto-discover the Agent’s name, description, and endpoint.
Description — Describe the Agent’s purpose.
Click Connect Agent to finish.
If your external Agent publishes an Agent Card, enter the Agent Card URL first. Runtype can fill in the name, description, and endpoint automatically.
Customize after creation
After you create an Agent from any template, open the Agent editor to fine-tune it.
Behavior tab — Edit the system prompt, adjust temperature, set max tool calls, and configure Agent Loop settings.
Capabilities tab — Add or remove Tools to expand what the Agent can do.
Templates set sensible defaults, so you can start testing right away and adjust as you go.
When to use a template vs. starting from scratch
Start with a template when you want a working foundation, want to learn how Flows or Agents are structured, or are building something common like support, research, data processing, or content generation.
Start from scratch when you have a unique workflow, want full control over every setting from the beginning, or already know exactly how you want to build your Flow or Agent.
