Welcome to your weekly AI Newsletter from AITechCircle!

I’m Building, 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:

• The Era of Agentic AI and blueprints to get started
• Generative AI Use cases in the Public Works Departments
• AI Weekly news and updates covering newly released LLMs
• Courses and events to attend

The Master class on AI

The Consumer Electronics Show (CES)​ , organized by the Consumer Technology Association, just concluded between January 6th and 10th. If you are working on or interested in generative AI, I recommend listening to this keynote from NVIDIA founder and CEO Jensen Huang.

This image caught my attention, and I am sharing the key to learning it. It also gives us a glimpse of the direction of current and future AI technologies.

• Perception AI helps us to understand text, speech, and images
• Generative AI helps us to generate text, speech, pictures, and videos
• Agentic AI can perceive, reason, plan, and act
• Physical AI can interact with the physical world

Perception AI: Understanding text, speech, images (and other sensory data)

Speech-to-text transcription, image classification, and entity extraction.

By extracting meaning from raw data, Perception AI allows computers to interpret and label the world around us.

Generative AI: Producing new content – text, images, audio, and video from learned patterns. Large Language Models (LLMs) generate text, and image diffusion models create artwork.

Generative AI goes beyond recognition and performs creative tasks, synthesizing outputs never seen before. Over the last two years, this has been the major driver behind AI’s growth.

Agentic AI: Perceiving, reasoning, planning, and acting autonomously.

An AI agent that can plan multiple steps and coordinate tasks (like multi-agent systems in robotics or autonomous decision-making tools).

Agentic AI adds a layer of autonomy on top of Perception and Generation. It takes inputs and reasons about them and executes actions to pursue objectives.

Physical AI involves interacting with the physical environment, such as robots or embedded AI systems that can manipulate objects or navigate spaces, autonomous drones, warehouse robots, and self-driving cars.

Physical AI merges computational intelligence with hardware capabilities to sense and act in the real world.

An example of 5 Blueprints from Nvidia for the Agentic AI:

Agentic AI orchestration coordinates multiple AI agents so they can function together effectively – a cornerstone for building robust enterprise AI solutions. NVIDIA’s orchestration partners have introduced blueprints integrating NVIDIA AI Enterprise, including NIM microservices and NeMo Retriever, to enhance retrieval accuracy and cut latency.

1. CrewAI employs Llama 3.3 70B NIM microservices and the NeMo Retriever embedding service for code documentation, keeping codebases comprehensive and easy to navigate.
2. Daily uses its open-source Pipecat framework, NVIDIA Riva, for speech recognition/text-to-speech, plus Llama 3.3 70B NIM microservices to deliver real-time conversational AI.
3. LangChain adds Llama 3.3 70B NIM microservices to its structured report generation blueprint so users can define a topic and outline, prompting the agent to research online and return a formatted report.
4. LlamaIndex uses NVIDIA NIM microservices and NeMo Retriever to create a blog creation assistant that automatically researches, outlines, and generates content (with source attribution).
5. Weights & Biases integrates W&B Weave into an AI virtual assistant blueprint featuring Llama 3.1 70B NIM microservices, enabling easier debugging, evaluation, and iteration for agentic AI deployments.

Nvidia Cosmos:

A platform designed to advance physical AI, Nvidia has unveiled a family of world foundation models (WFMs) – neural networks capable of predicting and generating physics-based video of a virtual environment’s future state.

These WFMs are as foundational as large language models. They use text, image, video, and motion data to create and simulate virtual worlds that accurately capture spatial relationships and physical interactions, paving the way for next-generation robotics and autonomous vehicles.

Weekly News & Updates…

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

1. ​North from Cohere: A secure AI workspace platform that aims to help teams build, refine, and deploy AI applications more efficiently. North focuses on delivering advanced developer tools, streamlined model training workflows, and robust data privacy features.
2. ​FineMath consists of 34B tokens and 54B tokens of mathematical educational content filtered from CommonCrawl.
3. ​Smolagents from Huggingface, a library to build agents

The Cloud: the backbone of the AI revolution

• Behind the Scenes: Using OCI Generative AI Agents to Improve Contextual Accuracy Link​
• Agentic AI: The next evolution of artificial intelligence link​

Generative AI Use Case of the Week:

Urban planning is evolving with the integration of Generative AI, offering innovative solutions to design challenges. Public Works Departments can now capitalize on the power of large language models (LLMs) to generate tailored urban design proposals for parks, housing layouts, and public spaces. These AI-driven suggestions streamline planning processes while aligning with community needs and sustainability goals.

To access the library of Gen AI Use cases, link here:

Chief AI Officer (CAIO) Corner:

Measuring Generative AI Maturity for organizations is vital, and this Framework provides a Clear Path to Growth.

​Tool for assessing Generative AI maturity within an organization

Favorite Tip Of The Week:

Here’s my favorite resource of the week.

​AI Agent Service Toolkit: The AI agent service is built with LangGraph, FastAPI, and Streamlit. It provides a template for quickly developing and running your agents using the LangGraph framework.

Potential of AI

Proptech Startup Raises $15M To Expand AI Insurance Tools Link​

Overall, AI startup funding in 2024 reached close to $314 billion compared to $304 billion in 2023, up around 3%, based on an analysis of Crunchbase data.​

Things to Know…

​SemiAnalysis has published the performance comparison between NVIDIA’s H100/H200 GPUs and AMD’s MI300X.

The Opportunity…

Podcast:

• This week’s Open Tech Talks episode 153 is “AI and Software Development: What Engineers Need to Know with Mayank Jindal,” He is a Software Development Engineer at Amazon.

Apple | Spotify | Amazon Music ​

Courses to attend:

• Course materials for class at UC Berkeley “Robots that Learn”​

Events:

• ​Oracle CloudWorld, KSA, Jan 19, 2025
• ​Oracle CloudWorld, Dubai, Jan 22, 2025
• ​AI in Dubai Conference, Jan 20 – 24, 2025, Dubai, UAE
• ​AI Everything GLOBAL, Feb 4-6, 2025, Dubai, UAE
• ​Web Summit, Feb 23-26, Qatar

Tech and Tools…

• AIHawk, the first Jobs Applier AI Agent: Your AI-powered job search assistant. Automate applications. link​
• ​Perplexica – An AI-powered search engine. It is an open-source AI-powered searching tool or an AI-powered search engine that goes deep into the internet to find answers

Data Sets…

• The 1000 Genomes Project is an international collaboration that has established the most detailed catalog of human genetic variation, including SNPs, structural variants, and their haplotype context.

Other Technology News

Want to stay updated on the latest information in the field of Information Technology? Here’s what you should know:

• Disruption Machine: How AI Is Reshaping Economies And Empires, published by Forbes​
• CES 2025: The 25 best products that impressed us the most reported by ZDNET​

And that’s a wrap!

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

I’d love to hear your thoughts. Please reply and let me know what you find most valuable this week. Your feedback means a lot.

Until next week,

Kashif Manzoor