Why understand AI Agents? Because investors are currently only focused on AI Agent companies, and cutting-edge companies like OpenAI are researching it.
The AI Agent is a direction for the application of AI to take root.
There's a joke in the entrepreneurial circle: I have a great idea, but I'm just short of a programmer. In the future, the AI Agent will be the 'programmer' that helps you realize your idea.
After reading this article, you will understand what an AI Agent is and how it works.
1. What is an AI Agent?
An AI Agent refers to an Artificial Intelligence Agent, which is an intelligent entity capable of perceiving the environment, making decisions, and executing actions.
For example, the AI Agent is like a little Xiao Ai, living in your phone or computer, with wisdom and observation ability.
When you say to it: 'Xiao Ai, I feel a bit unwell.'
It will sense that driving right now is not a good idea, so it conveniently calls the car to take you home, and it will arrive downstairs in 10 minutes. Hurry up and leave.
Then it actively generates a leave request for you, and with a nod, the leave request is sent to your leader.
It also kindly tells you that your home supply of ibuprofen and mineral water is running low, and it has already selected the products for you. Just give the command, and they will be delivered to your doorstep in 30 minutes.
It perceives that driving is not a good idea right now, so it conveniently calls a car for you to go home, which will arrive downstairs in 10 minutes. Hurry up and leave.
This is the result of a series of Agents working in collaboration.
2. With such excellence, how does it work?
A diagram to show you how an AI agent works. This diagram describes how an intelligent agent processes, analyzes, and responds to external information.
A bit unclear, right? Let me break it down.
The AI Agent is divided into 4 parts:
1. Perception
This is the first step of the process. AI establishes perception of the external world through sensors, cameras, microphones, etc.
Input: The perceived information is input into the system. In this case, the input includes: 'I feel a bit unwell,' my temperature, mental state, sleep duration, etc.
External Environment: The environment or context in which the system operates. For example, the situation of 'I feel a bit unwell' involves the weather, environment (such as whether there is pollen allergy), etc.
2. Information Processing (The Brain of the Agent)
It can be described as a general large model + numerous knowledge bases used to process information. It includes the following systems:
1) Information storage related
Memory System: Includes Storage and Memory, used to store long-term and short-term data.
For example, long-term data includes my basic information, hobbies, underlying health conditions, etc.;
Short-term data, such as there is only one bottle of mineral water left at home, can be deleted after the purchase.
Knowledge Base: Includes medical knowledge bases, product databases, etc., used to diagnose my current state and manage subsequent treatment and daily living needs.
2) The large model processes information.
Based on perceived information (input + Environment), memory, knowledge base, etc., it processes and derives conclusions (Decision Making): 'I am sick, and I have tested positive.'
3) Then develop the next step plan (Planning).
Action/Reasoning is based on the specific actions of its decisions but has not yet been implemented.
To help me write a leave request, buy medicine, buy water, call a car, etc.
3. Execution
Based on a series of dazzling operations by the Brain, a conclusion is drawn, and the next step plan is developed, which then requires execution (Action).
The large model itself cannot complete these tasks and needs to call external tools.
At this point, it will call third-party tools (Tools and Calling API), interacting with other apps through interfaces or applications to achieve the final effect.
4. Output
After execution, an output is needed to inform you of the results. For example, my Xiao Ai tells me: 'You have tested positive, and I have already prepared a leave request and called a car for you.'
This is the working principle of the AI Agent.
Overall, this system describes a simplified model showing how an AI agent starts from perceived information, undergoes a series of internal processing and decisions, and ultimately makes a response.
3. Summary
The AI Agent is one of the directions for future AI development (the other direction will be discussed in the next article).
It can be a personal assistant or a helper in your work, amplifying your capabilities, filling in your shortcomings, and making you a super individual.
