Welcome to your weekly AI Newsletter from AITechCircle!

I’m building and implementing AI solutions and sharing everything I learn along the way…

Check out the updates from this week! Please take a moment to share them with a friend or colleague who might benefit from these valuable insights!

Today at a Glance:

• Getting started with the OCI Gen AI RAG Agent
• Generative AI Use cases repository
• AI Weekly news and updates covering newly released LLMs
• Courses and events to attend

Creating your first Generative AI RAG Agent

As Agentic AI is the mainstream focus for 2025, if you are new to this topic, you can read an earlier post,’​​From Perception AI to Generative to Agentic to Physical AI​, ‘ which covers the fundamentals and the different stages of AI. This will enable you to understand the required concepts. Today, I will cover the overview of OCI Generative AI Agent Service, a managed service.

It combines LLMs with an intelligent retrieval system to create a contextually aware knowledge base for users’ queries and answers.

​

Three Types of Knowledge bases supported:

1. OCI Object storage files of pdf and text types as part of the managed service
2. Oracle Database 23ai vector search – bring your own option
3. OCI Search with OpenSearch ingested and indexed data – bring your own option

​

A retrieval-augmented generation (RAG) agent integrates retrieval capabilities with language generation to answer queries effectively. This AI combines factual precision from specific data sources with language models’ adaptability to craft contextually accurate and detailed responses. RAG agents are beneficial across various fields:

• Customer Support: Speeding up responses by pulling data from company databases.
• Legal Research: Assisting legal professionals by sourcing relevant case laws and precedents.
• Financial Analysis: Analyze extensive financial data and market reports.
• Educational Tutoring: Acting as tutors by providing educational content and resources.
• Technical Support: Offering guided troubleshooting assistance from technical documents.
• Supply Chain Management: Enhancing operational decisions by analyzing logistics data.
• Content Creation: Supporting content creators with data and draft suggestions.

Content Moderation:

Content Moderation in OCI Generative AI Agents is a feature that helps identify and block inappropriate content like toxicity, violence, and abuse in interactions with large language models. This tool classifies harmful content into four categories:

• Hate and Harassment: Includes identity attacks, insults, threats of violence, and sexual aggression.
• Self-Inflicted Harm: Covers self-harm and the promotion of eating disorders.
• Ideological Harm: Relates to extremism, terrorism, and organized crime.
• Exploitation: Encompasses scams and sexual abuse.

​

Let’s start exploring:

​

Open the navigation menu after logging into Oracle Cloud and select Analytics & AI. Under AI Services, select Generative AI Agents.

You will see the screen below with the three steps mentioned:

As a first step, you need to create the knowledge base, and for that, you need to have a bucked on the storage:

You should have a knowledge repository or documents in PDF or text format, which you will upload to this bucket. For example, I have created the ‘Organization HR booklet’ and ‘HR Leave policy.’

Now go back to Step 1 of creating the Knowledge Base and create it as per the screenshot below.

​

After setting the ‘Knowledge Base, ‘ go to Step 2: Create Agent.

Now, let’s interact with the Agent to test it and open the Chat from the left side of the menu.

You can see the results with the proper citation from where the agent is getting. Now, you can expose this agent via endpoint into your Digital Assistant/Chatbot or any GUI you already have.

If you are looking for a detailed architecture and how to use it in your ecosystem, you can read in-depth this article ‘GenAI RAG using OCI Digital Assistant and OCI Generative AI Agents’​

Data Handling in Generative AI Agent

In the OCI Generative AI Agents service, customer queries and related data are only used during a session and not stored afterward. This ensures that personal chat or knowledge base data is not used to train the model or enhance the service. The data from each session is handled privately without being retained or used for any purposes beyond immediate interaction.

Cost Estimates of Generative AI Agent:

Now, the key question is how to calculate the cost of this agent if you want to use it in your AI ecosystem. To answer this, head over to the Oracle Cost Estimator.

In the above example, I have chosen the following configurations:

• Request per month: 1000
• Average Number of total characters processed per request: 1,500
• Storage Capacity: 5GB
• Data Ingestion Jobs per month: 10
• Average Number of characters processed per data ingestion job: 50,000

The exported cost estimates are in Excel format below; you can download them to examine them.

​Gen AI Agent 16-02-2025, 7_09_02 PM.xlsx​

Call for Action:

Try it out yourself, log in to your Oracle Tenancy, and have fun learning how easy it is to get ready with your first Generative AI Agent.

​

Weekly News & Updates…

Last week’s AI breakthroughs marked another leap forward in the tech revolution.

1. The Anthropic Economic Index has been released to understand AI’s effects on labor markets and the economy over time. link​
2. Open AI roadmap for ChatGPT 4.5 and 5, few insights from Sam Altman​

The Cloud: the backbone of the AI revolution

• Agentic AI: A Game Changer for the Construction Industry, link​
• How Scaling Laws Drive Smarter, More Powerful AI link​

Generative AI Use Case of the Week:

Several Generative AI use cases are documented, and you can access the library of Gen AI Use cases. Link​

Favorite Tip Of The Week:

Here’s my favorite resource for the week.

OpenAI has released the best practices on using o-series models, primarily focusing on the three areas below:

• Differences between reasoning and non-reasoning models
• When to use reasoning models
• How to approach prompting reasoning models effectively

Things to Know…

​AI Action Summit was held in France to ensure the development of trusted, safe, and secure AI. The statement was signed by several countries, primarily focusing on the following priorities:

• Promoting AI accessibility to reduce digital divides;
• Ensuring AI is open, inclusive, transparent, ethical, safe, secure and trustworthy, taking into account international frameworks for all
• Making innovation in AI thrive by enabling conditions for its development and avoiding market concentration driving industrial recovery and development
• Encouraging AI deployment that positively shapes the future of work and labour markets and delivers opportunity for sustainable growth
• Making AI sustainable for people and the planet
• Reinforcing international cooperation to promote coordination in international governance

The Opportunity…

Podcast:

• This week’s Open Tech Talks episode 154 is “Generative AI Risks and Governance: What Business Leaders Need to Know with Terry Ziemniak.” he is a Fractional CISO and Partner at TechCXO

Apple | Spotify | Amazon Music ​

Courses to attend:

• ​Attention in Transformers: Concepts and Code in PyTorch

Events:

• ​SIFAMO AI SUMMIT, Tuesday 18, 2025, Dubai, UAE
• IDC Middle East CIO Summit 2025, Feb 19-20, Dubai, UAE
• ​Web Summit, Feb 23-26, Qatar
• ​Dubai AI Festival 2025, April 23-24, 2025, Dubai, UAE

Tech and Tools…

• ​Dify is an open-source LLM app development platform

And that’s a wrap!

Thank you, as always, for taking the time to read.

I would love to hear your thoughts. Please reply and share what you found most valuable this week. Your feedback means a lot to me.

Until next week,

Kashif Manzoor