Agentic Prompts

Karini AI’s Agent offers a comprehensive suite of capabilities to build advanced generative AI applications in complex production environments.

Using Karini AI’s prompt playground, you can create AI agents that integrate the reasoning capabilities of large language models (LLMs) with the ability to take actionable steps, creating a more sophisticated system that can understand and process information, evaluate situations, take appropriate actions, communicate responses, and track ongoing situations.

In the prompt playground, you can create an agent prompt by selecting the task type as Agent.

Key components of an Agent are:

• LLM (Large Language Model): LLM is responsible for generating traces, reasoning and actions for the task.

• Agent prompt: Agent prompt helps LLM in generating both reasoning traces and actions. Agent prompt may include variables, instructions for request handling and response processing.

• Tools: Tools enable the agent to interact with external systems to retrieve information and perform actions. In Karini AI, the Tools tab is organized into two sections:

  • Agent Tools: Tools configured directly for the agent (for example, internal connectors such as Catalog or Database). These tools are available for use during agent execution.

  • MCP Server: Tools provided through an MCP (Model Context Protocol) server. Connecting an MCP server allows you to expose and use externally hosted tools within the agent.

• Karini AI's agent prompt supports the following Agent tools:

  • Agent

  • Catalog

  • Database

  • Knowledgebase

  • Dataset (Vector Store)

  • Prompt (LLM)

  • REST API

  • KnowledgeGraph

  • Lambda

  • Amazon Q Retriever

  • Amazon Q Accessor Retriever

  • Amazon Bedrock Knowledge Base

  • Messaging

  • Browser Use

  • SAP Odata

  • Salesforce

  • Code Interpreter

Refer to the following video to create and test agent prompt.