Good day everyone! I've come across a task that I've never worked with before, maybe someone can guide me? tell me what to study, how it should look like. Thanks in advance to everyone☺️☺️☺️
Develop a command-line interface (CLI) application using Java or Kotlin that simulates an AI agent. This simulation is different from a real AI agent because it does not use LLM as a base - you will need to simulate this interaction manually. You can implement any agent you find interesting.
Requirements:
The input request is sent to the agent in text format. For example, "Get the temperature in London in an hour."
The agent also provides output in text format. For example, "In an hour, it will be 10°C in London."
The agent must be able to handle only one type of request. If you try to send it different types of requests, it will not be able to handle them. For example, the agent in the example above can handle requests like "Get the temperature at $location in $time", but it cannot handle requests like "Get the temperature at $location" or "Get the water temperature at $location in $time".
The agent must have some tools. For example, the agent can monitor its environment, analyze data, or perform some actions. Both data and actions can be synthetic if you want.
Recommendations:
You have full control over how your agent processes user input and whether it interacts with the user while processing a request.
Even though the agent will not use LLM, it will still need to somehow choose the appropriate tool while processing a request. This choice can be either non-deterministic or determined by certain conditions in the code.
Make sure you implement the agent, not the agent workflow.
Please attach a link to a public GitHub repository with the source code. Feel free to upload your work, even if it is not completely finished . #1051826284008853505