Podcast Image: Martin Sciarrillo from Microsoft

Martin Sciarrillo from Microsoft

From Cloud Pioneers to AI Trailblazers: Microsoft's Vision for Latin America's Tech Future

Host

Rod Rivera

🇬🇧 Chapter

Guest

Martin Tincho Sciarrillo

Director Data & AI for Hispanic South America, Microsoft

Martin Sciarrillo from Microsoft

This interview features Martín Ciarillo, the AI director at Microsoft for Latin America, conducted by Rod Rivera. Ciarillo discusses his professional background, the development of the GPT-RAG library, and the current state and future of AI implementation in Latin America.

The interview provides insights into the challenges and opportunities of AI implementation in Latin American businesses, emphasizing the role of cloud computing in enabling AI capabilities.

Takeaways

  1. GPT-RAG is a reference architecture for building efficient and secure AI solutions using Azure OpenAI.
  2. Latin America is accelerating its cloud adoption to leverage AI capabilities.
  3. Microsoft's Founders Hub program provides resources for SMEs to experiment with AI technologies.
  4. The quality of AI solutions depends on the quality of data provided.
  5. Latin America has great potential to contribute innovative AI solutions to global challenges.

Episode Transcript

Introduction and Professional Background

Rod Rivera: Today, I'm with Martín Ciarillo. He's the AI director at Microsoft for Latin America. Welcome, Martín.

Martín Ciarillo: Thank you for inviting me, Rod. It's a pleasure to be here.

Rod Rivera: The reason I became interested in your profile is because you created this GPT-RAG library that caught my attention. I wonder, how did you start? How is it that you're now directing AI for Microsoft Latin America? What has been your career path?

Martín Ciarillo: First, I want to clarify that we created the library as a team. I don't want to take individual credit for a collective effort, that's very important to me.

To tell you how I ended up in my current role, I should talk a bit about my journey at Microsoft and before. I come from a lifelong career in technology, with academic training in systems. I'm a systems analyst, and later I did professional training in management and other areas not directly related to technology.

I worked for pure technology companies like IBM, in the pre-cloud era, before the disruption of centralized computing resources appeared. Then I worked for EY (Ernst & Young), a consulting firm with a very broad technology organization. At that time, the cloud appears and I experience the transition from managing resources in our own data centers to starting to move workloads to the cloud.

We worked a lot with the cloud, learning through trial and error to understand where it was convenient and how to prioritize workload migration to the cloud. Three years ago, an opportunity arose to join Microsoft in a role focused on Argentina as CTO for Microsoft Argentina. I worked on the entire technology strategy for our large clients.

Role Change and Explosion of Generative AI

A little over six or seven months ago, the Argentine subsidiary was regionalized along with other subsidiaries. A new region is formed that we call Spanish South America (SSA), which includes Argentina, Colombia, Chile, Uruguay, and other Spanish-speaking countries in the region, excluding Brazil and Mexico.

My role changes and specializes in working specifically on data and artificial intelligence issues for all countries in this new large SSA subsidiary. This coincides with the great explosion of generative artificial intelligence. ChatGPT is released to the world in November-December of last year, with a vertiginous adoption: one million users in less than two months, quickly reaching 100 million.

In March of last year, the reinforcement of Microsoft's partnership with OpenAI is announced and we launch Azure OpenAI services, which are enterprise-scale OpenAI services.

Demand and Challenges in AI Implementation

The dizzying pace of what happened was enormous, even for those of us on the other side. With my team, we started to see a huge level of demand. All clients, both those who already consumed many of our services and those who didn't as much, expressed great interest in wanting to try and adopt this technology.

The common request was: "I want to be able to generate an experience similar to ChatGPT, but with my company's data". Use cases varied:

  • Training employees and setting up automated induction processes
  • Serving customers with specific industry and product knowledge
  • Creating conversational experiences fueled by each company's own information

When client information is involved, the necessary care and safeguards are completely different. We went through a period of initial turbulence with this explosive demand and then we started to organize ourselves.

Development of GPT-RAG

We realized that we needed to find a repetitive pattern aligned with best practices. Cloud architectures, not just in Azure, align with industry best practices (Well-Architected). So we set out to build a reference architecture for OpenAI-based solutions that would allow our clients to grow dynamically, using cloud resources, but also with adequate security conditions to protect data.

We started to refine the architecture, adding more components and consulting with experts within Microsoft. We asked ourselves: how can we secure these architectures? How can we add high availability components?

In a mass consumption solution, it may not be critical if you lose service availability for a few minutes. But when you're running a mission-critical service in a company, you need high availability. This is where the variables of running these types of solutions in the cloud play in favor, taking advantage of Azure OpenAI's presence in different regions of the world.

Key Features of Azure OpenAI

Martín Ciarillo: A fundamental point that we always clarify is that, unlike OpenAI, in Azure OpenAI all the information used belongs to each client and what you use within an instance in a private subscription is not used for any purpose outside of what the client defines. Not to retrain the model or anything. The information we put into an architecture for a specific subscription is used only for that.

Industries Interested in AI in Latin America

Rod Rivera: What types of companies have shown the most interest in Latin America? Are there some industries more interested than others?

Martín Ciarillo: The interest is generalized, I couldn't highlight one particular industry. Obviously, there are companies that have it easier. To use not only generative AI, but the most interesting features of Artificial Intelligence, you need your data to be in the cloud.

The prerequisite has to do with the maturity in the journey to the cloud that each client has. If the client has all their information on-premises in their own data center, they won't be able to capitalize on all the value of the solutions we have, at least from Azure.

In that aspect, Latin America is a bit behind in terms of the level of adoption of cloud-based solutions, especially cloud native. However, there has been a strong acceleration due to the interest in adopting these AI solutions.

We have clients from various sectors: public sector, retail, heavy industry, metallurgy, health... Typical use cases include training, induction processes, and customer service. Something these models have demonstrated is that they generate a very high level of empathy with those on the other side, creating a very fluid experience that people adopt naturally.

Recommendations for Small and Medium Enterprises

Rod Rivera: In Latin America, there are many small and medium-sized businesses, many family businesses that sometimes don't even have a dedicated systems department. For these types of companies that want to start offering some kind of chat service or automation, what do you recommend?

Martín Ciarillo: Although my focus is on large companies, Microsoft has thousands of clients within the small business segment. We have a program called Founders Hub, specifically designed to help that segment.

By registering your company in Founders Hub, you receive access to a number of very useful resources:

  • Azure resources
  • M365 resources (Office suite in the cloud)
  • Access to LinkedIn Premium
  • Azure credit (from 1,000to1,000 to 250,000)
  • Access to Azure OpenAI

In this way, SMEs can generate projects to help develop and digitize their business, gain time to market, and test these technologies.

Process for Large Companies

Rod Rivera: In your segment, which is large companies, if for example a Latin American multinational wants to introduce this, what is the first step?

Martín Ciarillo: In the large enterprise segment, they generally already have a technology sector. The profile of the data scientist, machine learning engineer, and AI administrator is being greatly redefined with the introduction of generative AI.

For Microsoft customers with a Unified contract, they can contact their team and we assign a Cloud Solution Architect who shows them how GPT-RAG works, tells them what use cases can be solved, and listens to their specific needs.

We also work a lot through partners. We have a huge network of Microsoft partners who are well trained in the use of our Generative AI technology and GPT-RAG.

Simple Explanation of GPT-RAG

Rod Rivera: If you have to explain to your family in a few words, in a simple way, what GPT-RAG is, what would you tell them?

Martín Ciarillo: GPT-RAG is like a blueprint for building cloud solutions, similar to how a house is designed. It's a pattern that serves to carry out ventures and solutions within the cloud, based on industry best practices.

It allows you to design solutions based on Azure OpenAI in the most efficient way possible. It's like having a ChatGPT-like experience, but with each company's own files and information, securely.

The most important thing is that the solutions produced with GPT-RAG will be as good as the information provided to them. If quality, well-formatted information is not given, the solution will be of low quality.

Trends and Future of AI in Latin America

Rod Rivera: What trends do you see? How do you see the future of AI in Latin America in the coming years?

Martín Ciarillo: In the next two years, I believe everything related to cloud movements will accelerate. The AI issue is an incentive for companies to understand that there are capabilities they will only be able to exploit 100% if their data is in the cloud.

I see that we will get closer to the level of maturity in the journey to the cloud seen in other countries of the world. In Latin America, there is an impressive number of startups and a huge level of creativity. Technology talent in Latin America is sought after worldwide.

I think we'll see more and more startups becoming successful companies beyond Latin America, relying on cloud adoption and the use of artificial intelligence. The cloud democratizes access to resources that were previously only available to large multinationals.

Final Message

Martín Ciarillo: I believe that navigating uncertainty is part of what we have to get used to in technology and in general. If one doesn't adapt to the disruption that's going to occur, it's very likely that they'll be left behind in a world that will increasingly require people who update themselves more quickly.

My recommendation is to explore and experiment. Today, the cloud and many solutions make technology within reach, whether for a large company or for a garage project of two people with a good idea.

The world is eager for solutions to big problems that we still have to solve as a society, and I believe Latin America has a lot to contribute in this regard.

Contact

Rod Rivera: Where can people find you if they want to contact you?

Martín Ciarillo: I'm quite active on LinkedIn. You can find me as Martín Ciarillo. There I share a lot about what I do with my team and what I see happening in the region that resonates with me. Microsoft's research area is super active, releasing very powerful open source projects, with code available on GitHub that any technology department or small project can implement.

Rod Rivera: Excellent, so we'll be looking for you on LinkedIn. Thank you very much for your time, Martín. It's been a pleasure.

Martín Ciarillo: It's been a pleasure, thank you for sharing our architecture and thank you for this space too, Rod. It's been great.

More from the Podcast

Post Image: E15: Alexa's $25B problem, AI upends marketing & search, sommelier's tech insights | Kyle Tsai

E15: Alexa's $25B problem, AI upends marketing & search, sommelier's tech insights | Kyle Tsai

Chris Rod Max dive into the challenges facing smart assistants like Alexa and Siri, exploring their limited use cases and struggles to monetize despite massive market penetration. Guest Kyle Tsai (ex-product manager, founder & sommelier) joins to discuss how AI and large language models could reinvent these platforms. The group debates the future of search and digital advertising in the age of AI copilots, potential opportunities for startups, and the need for marketers to adapt to a rapidly changing landscape. Plus, Kyle shares wine pairing recommendations for tech enthusiasts!

Rod Rivera

🇬🇧 Chapter

Post Image: Oscar Rovira from Mystic AI

Oscar Rovira from Mystic AI

In this captivating episode of AI Product Engineer, Oscar from Mystic AI takes us on a thrilling ride from the world of mechanical engineering to the cutting-edge realm of AI and machine learning. Discover how his passion for autonomous vehicles and artificial intelligence led him to England, where he honed his skills and embraced the power of Python.

Rod Rivera

🇬🇧 Chapter