Artificial intelligence is rapidly advancing, together with its potential for task automation. This paper discusses a new AI powered desktop application that uses the Large Language Model (LLM) engine to help its users automate tasks on their local Linux machines. The first component integrates a Chat Engine, which receives the user’s input and preserves the dialogue history. So, using pre-processed prompts, this engine describes task goals and functions that are available. And the LLM develops the final task in several stages based on these descriptions. The next stage in this process is the Response Formatter, whose task is to prepare instructions in response to the actions of AI Agent, so that they can be performed in order by the Execution Engine. A semantic level respecting the recursive architecture of the system enables it to decompose complex workflows into smaller subtasks and call on nonsense whenever it is needed. Based on seamless integration of generative AI and local machine control, this addresses the gap between natural language and task execution and hence become a useful tool for AI based Linux task automation.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Architectural Innovations in AI-Driven Automation: From Natural Language Prompts to Local Task Execution

  • Tejas Vaij,
  • Kriti Verma,
  • Vikas Verma,
  • Sonali D. Patil,
  • Azam Usmani

摘要

Artificial intelligence is rapidly advancing, together with its potential for task automation. This paper discusses a new AI powered desktop application that uses the Large Language Model (LLM) engine to help its users automate tasks on their local Linux machines. The first component integrates a Chat Engine, which receives the user’s input and preserves the dialogue history. So, using pre-processed prompts, this engine describes task goals and functions that are available. And the LLM develops the final task in several stages based on these descriptions. The next stage in this process is the Response Formatter, whose task is to prepare instructions in response to the actions of AI Agent, so that they can be performed in order by the Execution Engine. A semantic level respecting the recursive architecture of the system enables it to decompose complex workflows into smaller subtasks and call on nonsense whenever it is needed. Based on seamless integration of generative AI and local machine control, this addresses the gap between natural language and task execution and hence become a useful tool for AI based Linux task automation.