Skip to main content
The Lyzr Agent Builder is a conversational interface that allows you to move from an idea to a fully configured agent using natural language. Instead of manual configuration, you describe your requirements, and the builder handles the selection of models, tools, and personas.

​
Overview

Located directly on the Lyzr Studio Home Page, the Agent Builder serves as a “Fast Track” for agent creation. It bridges the gap between the granular control of the Studio Path and the code-first approach of the API Path.

​
The Conversational Build Process

​
1. Initiating the Build

On the Studio Home screen, use the central text area to describe the agent you want to create.
  • Example Prompt: “Build a Stock market analyst agent.”
  • Architect Integration: You can also click Explore Architect to see advanced templates and blueprints.

​
2. Feature Requirement Gathering

Once you submit your prompt, the Lyzr Agent Builder (Beta) opens a specialized chat interface. It analyzes your request and presents a series of checkboxes to refine the agent’s scope. For a “Stock Market Analyst,” the builder might suggest:
  • Market Research: Analyzing stocks and sectors.
  • Portfolio Analysis: Providing insights on existing portfolios.
  • Real-time Data: Fetching live fundamentals and prices.
  • News & Sentiment: Tracking market-moving news.
  • Technical Analysis: Chart patterns and indicators.
Tip: You can select multiple capabilities or use the text box at the bottom to say, “Actually, prioritize technical analysis over news.”

​
Integrated Governance & Testing

One of the unique advantages of using the Agent Builder is its native integration with Lyzr’s governance tools:

​
Simulation Engine

While you are building, the interface provides a preview of the Simulation Engine.
  • What it does: It automatically generates realistic test scenarios to stress-test your agent’s logic before you deploy it.
  • The Goal: Ensure the agent follows instructions perfectly and doesn’t hallucinate under complex queries.

​
Responsible AI & Approval Flows

Agents built via the conversational path are automatically mapped to your governance settings:
  • Responsible AI: Ensures the agent adheres to safety and bias guidelines.
  • Approval Flows: If configured, the newly built agent will be sent to an admin for review before it is enabled for production use.

​
Next Steps

Once the Builder completes the setup:
  1. Review in Studio: You can find the generated agent in the Agents or Voice tabs for further manual tweaking.
  2. Test in Simulation: Run a batch of simulations to see how it performs against “edge-case” user inputs.
  3. Deploy: Connect your agent to your preferred interface (Web, Twilio, or Slack).