Skip to main content

Custom Agent Overview

Custom Agents​

Creating a custom Agents​

JSONata with custom Agents​

What are Agents​

Agents and functions are specialized tools that allow the language model to perform specific tasks beyond generating text. These agents or functions are designed to extend the capabilities of the model by integrating it with external services or enabling it to execute specific operations that are useful in various applications. Here's an overview of what they are and their purpose:

1. Agents / Functions Overview​

  • Agents: Agents are essentially pre-built or custom functions that the model can invoke to perform specific actions. These actions can include things like retrieving real-time data, interacting with APIs, processing structured data, or even manipulating outputs in ways that go beyond natural language generation.
  • Functions: Functions refer to specific code snippets or operations that the model can call during its execution. These are often used to handle tasks that require deterministic processing, such as mathematical calculations, data formatting, or interacting with databases.

2. Purpose of Agents / Functions​

  • Extend Capabilities: Agents and functions allow the Reasoning Engine LLM to go beyond text generation and become more interactive and useful in real-world applications. For example, an agent can be created to fetch weather data, perform a stock market search, or send an email, making the model capable of performing a wide range of tasks.
  • Structured Interactions: They enable the model to handle structured data and make decisions based on it. This is particularly useful in scenarios where the model needs to interact with databases, spreadsheets, or any other structured data format.
  • Real-Time Information: Through agents,the Reasoning Engine LLM can access real-time information from various sources, such as checking flight status, looking up contact information in a CRM system, or fetching current news articles.
  • Custom Workflows: Users can create custom agents to automate specific workflows. For example, in a customer service chatbot, agents could handle tasks like booking appointments, checking order status, or even generating reports based on user queries.
  • Enhanced Decision Making: Functions can be used to perform calculations or logical operations that aid the model in making more informed decisions or generating more accurate responses. For example, calculating averages from a list of numbers or parsing JSON data for relevant information.

3. Examples of Agents / Functions​

  • Weather Agent: A function that retrieves current weather information based on a location query.
  • Email Sending Agent: An agent that allows the model to send emails with specified content to multiple recipients.
  • Database Query Agent: A function that can execute SQL queries against a database and return results for analysis.
  • Mathematical Operations: Functions to perform calculations like summing numbers, finding the mean, or even more complex mathematical operations.
  • Custom Data Parsing: Functions that parse and process JSON, XML, or other data formats to extract and manipulate information for use in responses.

In essence, agents and functions turn the LLM into a more powerful, interactive, and versatile tool capable of handling a wide array of tasks that go beyond traditional RAG