Executive Interview: Paola Santana, CEO and Founder of Social Glass

Paola Sanatana, CEO and founder of Social Glass
Company Takes a Data-Driven Approach to Governance; “We Can’t Change What We Can’t See” is the Social Glass Motto

Paola Santana is a lawyer, public procurement expert and tech entrepreneur creating the world’s next political system after democracy. She’s the founder and CEO of Social Glass, a software ecosystem using Artificial Intelligence to power high-performing governments. The company was recently named a top 10 e-Governance Solution Provider in govCIO Outlook. Previously, she co-founded a Silicon Valley startup that built a drone delivery platform, and co-established the Dominican Republic’s Constitutional Court. A Fulbright scholar and graduate from George Washington, Georgetown and Singularity University, she’s been featured as CNET Top 20 Latin in Tech, Forbes Top 50 Women of Power in Central America. She recently spoke for a few minutes with AI Trends Editor John P. Desmond about Social Glass, how it employs AI and how it fits into the political system.

AI Trends: Thank you for being with us today. You have described the Social Glass mission as to help create the world’s next political system after democracy. Could you elaborate on that? And where did the idea of the company Social Glass come from?

Paola Santana: For a while, we’ve been trying to connect the management of governance and governance operations with what type of political systems we live in. Especially here in the United States, we have a representative democracy and we continue iterating on that. I would even say we are putting Band-Aids on the system to improve what we have. And one of the key questions that we can pose ourselves is, should we continue improving the system we have today or should we take a couple of steps back, and redesign and rethink what system would work better for the society that we have today? 

We need to make leapfrog progress in transportation, education, health care, and other basic public services for the public good.

Social Glass offers the Micro product, an e-commerce marketplace for rules-compliant, fast, and paperless government micro-purchases.

As we rethink the systems, we have an opportunity to start observing the data, to start observing how governments operate today, and with that insight into their data and daily operations, to figure out if we need something different. And that something different, we don’t know what it will be called, and we don’t know what it looks like, we just know that we cannot make those decisions blindly because now we have the data. We have the opportunity to design something from scratch that works better for this society.

Creating a software ecosystem with Social Glass allows us to get insight into that data. The opportunity of creating Social Glass emerged two years ago when we were thinking about how to connect governments with technology tools that would allow us to make better decisions. We wanted to leverage all these exponential technologies like artificial intelligence and machine learning so we can see better what governments are doing and improve that decision-making and transform governments into high-performing entities. That was the original idea. We’ve been working on that and defining what the DNA of that is, and what a product that comes out of a software ecosystem like Social Glass is. We’re testing right now with our first product — MICRO — that is a marketplace for government purchases.

How is AI incorporated in your product?

In the hype of AI where everything has a little bit of AI or wants to be connected to some sort of rudimentary AI, it’s very important that we understand what AI can do for governments. The first thing is that developing general AI is not necessarily the goal here, to start. The main goal is, for example, using what we call narrow AI, systematizing very simple daily tasks and then being able to insert some type of machine intelligence to understand where there’s room for improvement.

For example, your Gmail inbox, spam inbox, has some sort of AI embedded on it, so it knows that if you’re receiving an email from someone that you’ve never been in touch with or something that looks like marketing material you are not interested in, it gets put in that spam folder. It’s not 100% accurate, but it’s pretty much there. 

That’s what we call narrow artificial intelligence, something that is not super futuristic. It’s something we can grasp. These are the type of little improvements that we think we can do today. With the types of marketplaces we’re creating now, we can advise a government agency on the best time to buy some new computers, for example. 

Lots of data analytics are employed. We examine historical data, and then apply narrow AI. At the end of the day, the decision is the humans to make. But we try to make suggestions that save money.

We need good data from governments. We need to capture data in a way that we can retrieve it, aggregate it and then start seeing insights. More data is not going to solve the problem. First we need to collect data in a way that all the data from every agency is meaningful and accurate. Once we have a substantial database of government data on how they are purchasing, how they make one decision over another decision, then we can start providing insights using AI.

Would you describe the AI that you use as machine learning or something else primarily?

It is narrow artificial intelligence. It’s a lot of predictive analytics. I couldn’t tell you that this is machine learning because the machine is not learning necessarily with new information. It is basically transforming the information it has, and using all the databases and other information that is in the markets to provide some very specific information to government. Machine learning would be something deeper, in our view. And these terms, as we go, continue changing. 

For us, machine learning would be having so much data that the software not only is providing insights, but is flagging that a poor decision is being made based, not only on historical data, but based on all the other decisions that other government agencies are currently making.

Machine learning is one step further. We’re not there yet. We’re just at the beginning of developing these predictive analytics models, and these very early algorithms of how to provide valuable insights to government with the information that we have today on their transactions.

How would you describe what stage the company is at in its life cycle today?

We are in a pre-seed mode, how we say in Silicon Valley, or in a pilot stage. We have developed our first product that is MICRO, this government marketplace. We are working with some government agencies here in California, on the local and state level, to test this in a private beta mode. We want to validate that MICRO is useful for them to do transactions under $10,000, what we call MICRO purchases, in a secure way where all the suppliers are curated and are verified so they can actually sell to government. And where they can see, in this one-stop shop, all their reviews and all the transactions they’ve done. And where they can manage some of their budget. So we are in this pilot mode, where we are validating everything we’ve been observing, and we’re discussing it with the government agencies. We want to see how they are using the product so we can improve it and then do a public launch of the product by the end of the summer. That’s the goal.

We are seeing very big changes on the procurement market. The current administration has increased the micro-purchase threshold from $3,500 to $10,000. The federal government has mandated that goods under the micro-purchase threshold be purchased through e-commerce marketplaces by 2020. The GSA is working very hard to implement this mandate.

So we are working to design a product for the government user that has a specific need, without them having to do the work of aggregating all the purchase data. 

The US has almost 90,000 different government units doing over 160 million transactions a year. Many of these transactions are micro-purchases. If we can optimize these transactions, we can help purchasers maximize their budget. We need to make sure that we are providing them value on the small scale first. We are doing the pilot to make sure that we can claim that we helped on 100 transactions, so we can help on 160 million transactions.

Is it realistic to take a data-driven approach in a political context?

Not only is it realistic but it’s the next step for making better decisions on behalf of everyone. There have always been very close ties between politics and governance; we agree that politics defines if we’re going to have more budget for health care or less budget for healthcare. But public healthcare is on the agenda. So how can we have public healthcare functioning in a high-performing way? We’re always comparing the way government operates with the way a private enterprise operates. The question is, how can we insert some sense of urgency into how we provide public services? That’s number one. 

Number two, how can we improve those public services, not through perception, not through what we read on the press, not through what one politician said or Congress said, but through data. And the more data points we have on how things are being managed, the better. The more insight we have into the data of what is actually happening, the better output we will get. The Social Glass motto is, “You can’t change what you can’t see.”

Can you tell us about the pilot projects, how they’re being applied in the real world as much as you can, if you’re not able to say specifically what agencies you’re working with?

We’re working with two local agencies in California in a low-risk environment, meaning that we are not changing any process that they currently have. We are not asking them to stop using any piece of software that they already have. What we’re doing is connecting many of the current practices that they have and much of the different software that they have, and putting everything under one umbrella and testing that with some very specific users inside their agency. We want to see if that simplifies their decision-making or even eliminates some steps that are not needed.

The main problem with micro-purchases is the purchases are doing in many different ways, even in different departments inside the same agency. We see room for improvement.

I’m not going to name the agencies, but in some agencies here, they purchase IT goods like software and subscriptions through one department, goods through another department and services through another department. 

We are thinking we can put all of that under one umbrella, and guarantee that every supplier in our marketplace is providing government with their best available price. And they will not find that price anywhere else; they will find the best price under MICRO. And then make sure that once those services are provided, these government purchasers can see all their transactions in one place. They can see their analytics and they can also provide insights to their financial department. 

We want to show at the end of the pilot that we have provided these government agencies with something of value, that they can then show to other cities and government agencies. Then we can start creating that ecosystem where governments are really saving time and money because we are scaling the solution in California, then in other states.

That’s as much as I can say about the pilot program. Any government agency can apply to our pilot program by, you know, by telling us what their specific need is. Because we’re a startup, we would analyze if we have bandwidth to really support them. But we don’t need a lot, we only need access to one of the persons in charge of doing purchases and then we take it from there. We analyze the whole process. We analyze the opportunity for improvement. And we start providing them with insights into how they’re doing their processes and then providing an insight report by the end of the month. So that’s what the pilot program is about right now.

Do you have a timeframe yet for building a scalable enterprise product?

By the end of this summer, we will have the public launch of MICRO that is now being tested under a pilot program. And then, by the end of the year, we expect to have at least 10,000 transactions in the system from California. That sounds like a small number compared to all that government purchases, but it will allow us to see opportunities. We are planning to start expanding outside of California in Q1 of 2020. 

Learn more at Social Glass.