🎆
Real-World Agentic Applications
Status Completed
Tags Agents LLM genAI
https://medium.com/@cobusgreyling/real-world-agentic-
Link
applications-c527d24405a2
Author Cobus Greyling
End
@November 6, 2024
Date
Source Medium
Start
@November 6, 2024
Date
Time 5
Real-World Agentic Applications
Real-World Agentic Applications 1
Introduction
basic architecture of AI agents is well-established
agents break down complex tasks in smaller sub-tasks, creating an ad-hoc
chain of actions
at each step, the agent follows a process of “thought, action, and observation”
to address or complete the sub-tasks
this iterative cycle continues until the final conclusion or answer is reached
this dynamic and flexible approach allows AI agents to handle intricate
challenges by systematically tackling each component of a task
see visual to link to rest of article
Application Environments
AI agents need to operate within an environment which can be digital spaces
like mobile operation systems or the web
these are the same environments users already engage with, positioning AI
agents as an extension of the user → agents can integrate seamlessly into
user workflows
Real-World Agentic Applications 2
Training Data
proposed architecture begins with training data and highlights how data
sources have evolved
shift moves beyond simple data chunking, adopting more advanced
techniques
by structuring data this way, agents can better understand and process
complex relationships within datasets, enhancing their ability to navigate
and perform tasks more effectively
few-shot fine-tuning allows the model to adapt its performance to a niche or
specialised task without requiring extensive new data, while preserving
general knowledge acquired during training
Foundation Model
foundation model serves as core backbone for AI agents, providing the
necessary infrastructure for various tasks
by combining the foundational capabilities of large models with domain-
specific functionality, these agents can effectively navigate and perform
complex tasks in diverse environments
Fine-Tuned Models
model orchestration is getting more and more attention
by combining large models and few-shot fine-tuning, you get fine-tuned
models, adapted to a specific task
Environmental Realities
refer to the cognitive abilities and the environments in which AI agents can
operate
currently: agents exist primarily as digital entities, but advancements in
robotics, sensors, and AI may enable agents to interact with the physical world
AI Agent Architecture
Real-World Agentic Applications 3
basic AI architecture emphasises the importance of the environment in which
the agent operates
for an agent to function effectively within its environment, it requires cognitive
components
Finally
matrix below outlines the dual actions and environments within which AI
agents operate
Real-World Agentic Applications 4
agent’s capabilities are split across virtual and physical environments and
between manipulation and intent actions
virtual environments: AI agents decompose tasks, reason, and navigate,
such as selecting tools to complete tasks
physical environments: AI agents assist in real-world contexts, handling
tasks like manufacturing
intent action: agents detect user intentions through inputs, while personal
assistants with physical with physical autonomy extend these capabilities
to real-world interactions
Real-World Agentic Applications 5
