The Best Fluffy Pancakes recipe you will fall in love with. Full of tips and tricks to help you make the best pancakes.
Architecture is the way and compositions on which AI agents have been designed. Some AI Agents such as LLMs have a different architecture because they need a larger memory while those with reflexive tasks such as preventing DDoS attacks have a lower memory but higher reactivity, and hence are differently composed.
Based on Architecture
Based on what they do in the real world, models can be divided into 5 categories:
- Simple Reflex Agents
- Model-Based Agents
- Goal-based Agents
- Utility-based Agents
- Learning Agents
1. Simple Reflex Agents
These agents are created to react to a certain event in an environment. For example, an AI agent that blocks a specific kind of URL from appearing on your search results when you search for something.
These agents do not have a long-memory but just the rules and regulations that help them achieve their goals.
Another such agent could be the one used to block DDoS attacks by stopping fake user traffic.
2. Model-Based Agents
These agents have AI Large Language models at their core. They are derived from some AI model like “Operator” is derived from ChatGPT.
They also have a vivid representation of the external world inside them. For example, barring a few months of lag, ChatGPT’s Operator can tell you most of the things about this world with high accuracy.
3. Goal-based Agents
Goal-based agents are created to achieve a certain goal.
For example, a goal based agent could be created to fill the DasAI website with glossary terms related to AI. At the end of the goal, these agents will simply be terminated.
Another example of such AI agents would be driverless cars, which must get a person safely and comfortably from point A to B without malfunctioning.
4. Utility-based Agents
These agents are also sometimes called as Vertical AI because they are specifically designed to perform a special type of task.
These agents achieve all kinds of tasks based on a single nature.
For example, an AI agent that plans long-distance journeys based on the preference set by the user.
Another example is a financial agent that helps people with:
- designing portfolios
- make savings plan
- create a DB of top stocks and their prices
- analyze cryptocurrencies based on growth prospects
5. Learning Agents
Learning Agents are those which have been designed to scrape information from the internet by observing and learning things on the web. These agents simply gather information, compile them and deliver as and when required.
They self-improve based on the new learnings they absorb periodically.
Examples of these AI agents would be self-updating GPTs like GrokAI.
