Latest Posts Matías Battaglia

Emily Johnson

-Mar 12, 2026, 5:24 PM

In this post, I’ll kick off a series of posts where I’ll be diving deep into Retrieval Augmented Generation (RAG), covering both the theory and building one from scratch. I’ll start by explaining what it is, what its main components are, and building a simple implementation. Over the next the few posts in the series, I’ll explain additional RAG techniques and iterate over the implementation. I will try to keep a very simple approach and do my best to always provide practical advice. Retrieval Augmented Generation (RAG) is a technique that dynamically fetches relevant information from external knowledge bases and adds it as part of the context sent to an LLM. The underlying goal is for the LLM to use this added information to generate a response instead of purely using its own parametric data (the data that it learnt during its training), increasing the...

At its core, a step-by-step RAG workflow looks like: Even as AI continues to move forward and more organizations release increasingly capable LLMs, there are a number of underlying limitations in the technology that remain yet to be solved. Some of these are significant and limit the real applicability of LLMs for many use cases. Some examples are the lack of real understanding of the physical world, their non-deterministic behavior, their lack of long-term memory, etc. While some of these are inherent to the technology and cannot be worked around, others can be addressed with the right architectural patterns. I want to talk about how LinkedIn is no longer what it was before, how the use of IA has made every post bland and trendy, how people no longer share knowledge or useful...

I want to talk about how unprofessional the Professional Network app has became. I log in and I see a myriad of shallow, contentless posts about AI, about what the future is, what the new trend is and how you need to keep up in order to... Guess what, I disagree, I don't think you need to be on top of the latest trend to be a great candidate or a good worker. I'm going to just limit to my narrow area of expertise (if you can call it that way): On a daily basis I see tenths of posts about a new approach to quality, or... And the most annoying thing to me is that no one is challenging the content being shared, either no one dares to speak up, or no one cares about it. /rant off

Tu posteo da para un debate de 5 horas! Muchos "desafiamos" sin comentar ni likear, para no darle más protagonismo a los influencers del bla bla. Me parece serio y preocupante que muchos piensen que la visibilidad acá es lo que determina tu calidad laboral, profesional, tu capacidad de aprender, etc. No es más que una nueva versión de la teoría del rumor. De igual manera, creo que podemos cambiarlo, con IA, con oportunidades, con crear contenidos saludables. Se pueden imponer otros #.

you forgot the "/rant on" tag at the beginning (lol) Linkedin isn't usefull for anything, heck it doesn't even help landing a job anymore Here are some of the talks and presentations I’ve given at conferences, meetups, and other events. I’m always interested in speaking at conferences, meetups, and other events. If you’d like me to speak at your event, please reach out on my Linkedin! It was an absolute blast being part of the GenAI Demo arena at the AWS Summit Madrid!

Sharing the stage with Sergio Garcés Vitale and showcasing the transformative power of GenAI through live demos was truly rewarding. I'm thrilled that these demos are now available for customers to witness the 'Art of the Possible' firsthand. It's been a month since the AWS Summit Madrid, and we're still excited by the response to our *GenAI Demo arena*. Our goal was to showcase the "Art of the Possible" with real-time demos of how customers and partners are transforming their businesses with GenAI. We had the privilege of presenting 20 live demos to 4,508 attendees, highlighting the power and versatility of GenAI across various industries. From ComfyUI's engaging visual content creation tool to PhenoXtractor's medical information transformation, from an innovative education platform providing concise summaries and instructional flashcards to the potential of GenAI in healthcare data management.

And this was just the beginning. All these demos are now available for our customers, and we're eager to explore how GenAI can revolutionize your business. Let us know if you'd like us to demonstrate the transformative power of GenAI tailored to your specific needs. And thank you Diego Rodríguez Herrero for being the best partner in crime. And all Marketing team for trusting us to make this real, Diana T., CAROLINA GÓMEZ PARDAL, Valery De Sousa and Arancha Sanz Manzano. #genai #generativeai #awssummit #aws

There’s a lot of hype circulating about using agents to solve pretty much every use case under the sun. There’s extensive discussion and content about multi-agent systems, orchestration and coordination layers, and much more. While this approach will certainly be the right solution for some advanced use cases (and I’ll share my opinion on when those solutions make sense later), this current trend makes it look like the... Complexity creates barriers for organizations and individuals taking their first steps in the AI world. It’s my duty as a technologist to stand against unnecessary complexity and to preach for simpler, easier-to-understand systems and patterns. In their publication on agentic AI, Anthropic analyzed successful real-world implementations and found:

Consistently, the most successful implementations weren’t using complex frameworks or specialized libraries. Instead, they were building with simple, composable patterns.

Latest Posts Matías Battaglia

People Also Search

In This Post, I’ll Kick Off A Series Of Posts

At Its Core, A Step-by-step RAG Workflow Looks Like: Even

I Want To Talk About How Unprofessional The Professional Network

Tu Posteo Da Para Un Debate De 5 Horas! Muchos

You Forgot The "/rant On" Tag At The Beginning (lol)