Building Business-Ready Generative AI Systems : Build Human-Centered Generative AI Systems with Agents, Memory, and LLMs for Enterprise.

Format:

Book

Author/Creator:

Rothman, Denis.

Language:

English

Subjects (All):

Artificial intelligence--Computer programs.

Artificial intelligence.

Artificial intelligence--Business applications.

Physical Description:

1 online resource (0 pages)

Edition:

1st ed.

Place of Publication:

Birmingham : Packt Publishing, Limited, 2025.

Summary:

"Unlock the skills to design and implement generative AI systems tailored for enterprise applications with ""Building Business-Ready Generative AI Systems"". This book covers advanced concepts like AI controller architectures with memory, multimodal reasoning, and secure integration of LLMs and other AI models. You'll gain practical knowledge to bring artificial intelligence solutions to the forefront of your business. What this Book will help me do Design and develop AI controller architectures with advanced memory retention. Create and integrate multimodal reasoning and image generation capabilities. Implement Generative AI systems using cutting-edge models like OpenAI LLMs. Apply Chain-of-Thought (CoT) orchestration for enterprise-level problem solving. Build secure, scalable AI systems for business domains including marketing, production and logistics. Author(s) Denis Rothman is a seasoned author and AI practitioner with extensive experience in developing intelligent systems for enterprise solutions. With a background in computer science and a career spanning over decades in AI development, Denis brings both the technical depth and practical insights necessary to guide readers in building advanced AI solutions. His books are known for providing clear, actionable content. Who is it for? This book is ideal for AI and machine learning practitioners aiming to develop enterprise-grade Generative AI systems. Whether you're a software architect, data scientist, or business professional expanding into AI, this guide offers actionable insights to broaden your expertise and implement effective solutions. A basic understanding of AI concepts is recommended to get the most out of this book.".

Contents:

Cover

Title Page

Copyright

Dedication

Contributors

Table of Contents

Preface

Your Book Comes with Exclusive Perks - Here's How to Unlock Them

Chapter 1: Defining a Business-Ready Generative AI System

Components of a business-ready GenAISys

AI controllers

Model-agnostic approach to generative AI

Building the memory of a GenAISys

RAG as an agentic multifunction co-orchestrator

Human roles

GenAISys implementation and governance teams

GenAISys RACI

Business opportunities and scope

Hybrid approach

Key characteristics

Use case examples

Small scope and scale

Full-scale GenAISys

Contextual awareness and memory retention

Setting up the environment

Downloading OpenAI resources

1. Stateless and memoryless session

Semantic query

Episodic query with a semantic undertone

Stateless and memoryless verification

2. Short-term memory session

3. Long-term memory of multiple sessions

4. Long-term memory of multiple cross-topic sessions

Summary

Questions

References

Further reading

Chapter 2: Building the Generative AI Controller

Architecture of the AI controller

Conversational AI agent

Conversational AI agent workflow

Starting the initial conversation

The full-turn conversation loop

Running the conversational AI agent

Next steps

AI controller orchestrator

Understanding the intent functionality

From T5 to GPT models

Corpus of Linguistic Acceptability (CoLA)

Translation task

Semantic Textual Similarity Benchmark (STSB)

Summarization

Implementing the orchestrator for instruction selection

Selecting a scenario

Defining task/instruction scenarios.

Performing intent recognition and scenario selection

Running scenarios with the generative AI agent

Sentiment analysis

Semantic analysis

Chapter 3: Integrating Dynamic RAG into the GenAISys

Architecting RAG for dynamic retrieval

Scenario-driven task execution

Hybrid retrieval and CoT

Building a dynamic Pinecone index

Installing Pinecone

Initializing the Pinecone API key

Processing data

Data loading and chunking

Embedding the dataset

Creating the Pinecone index

Upserting instruction scenarios into the index

Upserting classical data into the index

Querying the Pinecone index

Querying functions

Querying the vector store and returning results

Processing the queries

Retrieval queries

Chapter 4: Building the AI Controller Orchestration Interface

Architecture of an event-driven GenAISys interface

Building the processes of an event-driven GenAISys interface

1. Start

2. Initialize widgets

3. Display the UI

4. Input box event

5. chat(user_message) function

6. If 'exit' is chosen

7. If user(s) continue the conversation

8. Generate bot response

9. Update display

Conversational agent

Multi-user, multi-turn GenAISys session

A session with two users

The interactive conversation

Loading and displaying the conversation

Loading and summarizing the conversation

Multi-user session

Semantic and sentiment analysis

RAG for episodic memory retrieval

Generative AI agent for ideation

Dialogue without an AI conversational agent

Loading, displaying, and summarizing the conversation

Further reading.

Chapter 5: Adding Multimodal, Multifunctional Reasoning with Chain of Thought

Enhancing the event-driven GenAISys interface

IPython interface and AI agent enhancements

Layer 1: IPython interface

Layer 2: AI agent

Layer 3: Functions

OpenAI

Initializing gTTS, machine learning, and CoT

Image generation and analysis

Image generation

Image analysis

Reasoning with CoT

CoT in GenAISys versus traditional software sequences

Cognitive flow of CoT reasoning

Start

Step 1: ML-baseline

Step 2: Suggest activities

Step 3: Generate image

Step 4: Analyze image

End

Running CoT reasoning from a user perspective

Chapter 6: Reasoning E-Marketing AI Agents

Designing the consumer GenAISys memory agent

Consumer-memory agent use case

Defining memory structures

Enhancing the architecture of the GenAISys

Building the consumer memory agent

The dataset: Hotel reviews

Step 1: Memory and sentiment analysis

Designing a complex system message for Step 1

Running the memory analysis

Step 2: Extract sentiment scores

Step 3: Statistics

Step 4: Content creation

Step 5: Creating an image

Step 6: Creating a custom message

GenAISys interface: From complexity to simplicity

Adding the CoT widget

Enhancing the AI agent

Generalizing the GenAISys capabilities

Chapter 7: Enhancing the GenAISys with DeepSeek

Balancing model evolution with project needs

DeepSeek-V3, DeepSeek-V1, and R1-Distill-Llama: Overview

Getting started with DeepSeek-R1-Distill-Llama-8B

Setting up the DeepSeek Hugging Face environment

Downloading DeepSeek

Running a DeepSeek-R1-Distill-Llama-8B session

Integrating DeepSeek-R1-Distill-Llama-8B.

Implementing the handler selection mechanism as an orchestrator of the GenAISys

What is a handler?

Why is a handler better than a traditional if...then list?

1. IPython interface

File management

2. Handler selection mechanism

3. Handler registry

Pinecone/RAG handler

Reasoning handler

Analysis handler

Generation handler

Image handler

Fallback memory handler

4. AI functions

RAG

Sentiment analysis (genaisys)

Semantic analysis (genaisys)

Data retrieval (data01)

Chain of thought

Analysis (memory)

Generation

Creating an image

Fallback handler (memory-based)

Chapter 8: GenAISys for Trajectory Simulation and Prediction

Trajectory simulations and predictions

Challenges in large-scale mobility forecasting

From traditional models to LLMs

Key contributions of the paper

Reformulating trajectory prediction as a Q&amp

A

Instruction tuning for domain adaptation

Handling missing data

Building the trajectory simulation and prediction function

Creating the trajectory simulation

Visualizing the trajectory simulator

Output of the simulation function

Creating the mobility orchestrator

Preparing prediction instructions and the OpenAI function

Message preparation

Fitting the messages together

Implementing the messages into the OpenAI API function

Trajectory simulation, analysis, and prediction

Adding mobility intelligence to the GenAISys

IPython interface

Creating the option in instruct_selector

Handling the "mobility" value in update_display()

handle_submission() logic

Handler selection mechanism

AI functions

Running the mobility-enhanced GenAISys

Production-delivery verification scenario

Fire disaster scenario

Chapter 9: Upgrading the GenAISys with Data Security and Moderation for Customer Service

Enhancing the GenAISys

Adding a security function to the handler selection mechanism

Implementing the security function

Handler selection mechanism interactions

Implementing the moderation function

Building the data security function

Populating the Pinecone index

Running security checks

Building a weather forecast component

Setting up the OpenWeather environment

Adding a weather widget to the interface

Adding a handle to the handler registry

Adding the weather forecast function to AI functions

Running the GenAISys

A multi-user, cross-domain, and multimodal dialogue

Chapter 10: Presenting Your Business-Ready Generative AI System

Designing the presentation of the GenAISys

Building a flexible HTML interface

1. Presenting the core GenAISys

2. Presenting the vector store

3. Human-centric approach to KPIs

ROI through growth

Adding a real-time KPI to the GenAISys web interface

4. Integration: Platforms and frameworks

Showcasing advanced frameworks: A MAS

Strategic integration options for the MAS

5. Security and privacy

6. Customization

7. GenAISys resources (RACI)

Answers

Other Books You May Enjoy

Index.

Notes:

Description based on publisher supplied metadata and other sources.

ISBN:

1-83702-068-X

OCLC:

1528957024

Publisher Number:

CIPO000255927