In the previous article, we explored the difference between LAM (Large Action Model) and LLM (Large Language Model), which have been frequently mentioned alongside the popularity of ChatGPT. Today, we’ll take a closer look at how LAM (Large Action Model) is connected to the AI Agent, which is often mentioned when explaining LAM.

For those interested in understanding more about the difference between LAM and LLM, you can check out the link to the second article here.

LLM vs LAM: What is the Difference Between Language Models and Action Models

‍

LAM and AI Agents

As explained in the previous article, LAM has the ability to predict and execute actions based on various inputs (images, videos, etc.) beyond text. In other words, when LAM is presented with typical or unexpected situations, it determines the optimal action to take in order to achieve the best outcome..

On the other hand, an AI Agent is an individual entity that directly carries out tasks based on the given environment and information in order to achieve a user-defined goal. In essence, an AI Agent acts on the plan set by LAM and performs the actual tasks.

LAM is the Brain, AI Agent is the Body

To simplify, LAM can be considered the "brain" while the AI Agent is the "body" that performs the actions.

A helpful comparison to understand the relationship between LAM and the AI Agent is the construction of a building.

In the process of building a structure, the architect (LAM) creates the design and oversees the entire process from the foundation to the completion of the building.However, despite the architect’s detailed plans, people (AI Agents) are needed to physically carry out the work, such as laying bricks, painting, installing electrical systems, and handling civil engineering tasks.

Thus, LAM is like the architect, responsible for overall design and decision-making, while the AI Agent performs the practical work on the construction site. While the AI Agent collects data and carries out tasks that align with the user's intent and goals, LAM ensures that everything progresses optimally toward the desired outcome.

Interaction Between LAM and AI Agents

To break down the comparison further, the interaction between LAM and AI Agents can be explained step by step as follows:

1. User Request: When a construction project is initiated, the architect (LAM) formulates an outline for the building design and gathers relevant data through collaborators (AI Agents).
2. Request Analysis and Transfer to LAM: The collaborators (AI Agents) gather the necessary data and deliver it to the architect (LAM), who analyzes the information.
3. LAM Decision-Making and Planning: The architect (LAM) uses the analyzed data to create the building design and construction plans. They also prioritize tasks and define workflows.
4. AI Agent Task Execution: Following the architect's design and construction plan, the AI Agents (workers) perform the actual work on-site, such as laying the foundation, building walls, and conducting electrical work.
5. Progress Updates and Optimization: As the work progresses according to the design and construction plans, unexpected situations may arise. The AI Agents report these to the architect (LAM), who identifies the best course of action to ensure the project aligns with the user's goal. This process of reporting and adjusting continues until the building is completed (goal achieved).

Conclusion

Ultimately, LAM and AI Agents complement each other in a collaborative relationship.

LAM is responsible for developing the overall strategy and plan, while AI Agents execute that plan to achieve tangible results aligned with the user’s goals. This interaction is essential for achieving optimal outcomes, and when LAM and AI Agents work together in an integrated way, their efficiency and effectiveness are maximized. T

his collaborative structure plays a crucial role not only in task execution but also in adapting to and learning from various changes to reach the best possible outcome for the user-defined goal.

‍

If you're curious about the LAM-based AI Agent that Enhance is preparing, you can check out the article linked below.

The Coming Era of AI Agents... How  Enhans Answers the Future of Digital Commerce?