AI's Invisible Hand is Fixing Government Spending's $16 Trillion Blind Spot

SAN FRANCISCO, CA – July 27, 2026

Governments worldwide spend an estimated $16 trillion annually on goods and services, yet a staggering portion of that activity is obscured by a fundamental, almost mundane, problem: they don’t have a consistent way of describing what they buy. A single model of laptop might be recorded under dozens of different names, prices, and product codes across various agencies. This data chaos creates a massive blind spot, making it nearly impossible to track spending, compare prices, or ensure public funds are used effectively. Now, a GovTech firm from Silicon Valley, in a quiet but transformative partnership with the government of Puerto Rico, has deployed an AI-driven solution that operates invisibly to fix the problem at its source.

Glass, a company specializing in government procurement technology, today announced its Master Catalog, an AI system developed with Puerto Rico’s Administración de Servicios Generales (ASG). The system doesn't just clean up messy procurement data; it creates a permanent, living, and standardized record of every single item a government purchases. The results from the Puerto Rico pilot are a red flag for the status quo and a blueprint for the future of public financial management.

The Anatomy of a Multi-Trillion-Dollar Blind Spot

For years, Puerto Rico’s 105 government agencies operated in procurement silos. A simple box of pens purchased by the Department of Education might be logged as "PEN, BALLPOINT, BLK," while the Department of Health recorded the same item as "WRITING INSTRUMENT, INK, 10-PACK." This wasn't a failure of data collection; it was a failure of structure. Without a shared language, the millions of data points generated by public purchasing became a tangled web of inconsistencies.

This challenge is not unique to Puerto Rico. It is the default state for most public entities. The consequences are profound. When you cannot reliably compare what you bought, you cannot compare prices to ensure you got the best value. You cannot aggregate purchasing across agencies to leverage bulk discounts. Most critically, you cannot provide taxpayers or oversight bodies with a clear, accurate picture of how their money was spent. This data fragmentation has long been a frustrating operational hurdle, but in an era of tightening budgets and demands for greater accountability, it has become a critical financial liability.

Puerto Rico’s government recognized this as a core obstacle to its broader reform efforts, which were codified in the 2019 law known as Act 73. The legislation aimed to centralize and bring efficiency to a historically fragmented procurement system. To achieve that goal, they needed more than a new policy; they needed new infrastructure.

An AI-Powered Blueprint for Clarity

Enter Glass. Instead of building another dashboard to sit atop the messy data, the company went to the foundational layer. The Glass Master Catalog ingests historical purchasing records—in Puerto Rico's case, nearly 300,000 raw purchase orders—and puts them through a sophisticated, multi-stage AI process.

First, deduplication algorithms sweep through the data, identifying and merging redundant entries for the same product. From there, AI agents take over, normalizing product descriptions, correcting inconsistencies, and enriching each entry with complete information. A second layer of AI then performs the crucial task of classification, assigning each of the 110,318 validated items a standardized NIGP (National Institute of Governmental Purchasing) code with an average confidence rate of 99.20%.

The efficiency gains are staggering. A task that once took a procurement officer around ten minutes of manual review per item now takes the AI an average of 48 seconds—a 91.7% reduction in processing time. The cost is equally compelling: approximately $0.002 per item classified, a fraction of what equivalent human review would demand.

"Our goal was never simply to 'use AI'—it was to design an intelligent system that works invisibly in the background, reduces manual effort, and preserves the workflows buyers already know," said Anthony Rivas, Chief Revenue Officer at Glass. The real breakthrough, he explained, is not the one-time cleanup but the system's ability to stay alive. A second AI system works continuously behind the scenes, integrated with the government's existing ERP environment. When an agency buys a product, the AI instantly recognizes it, matches it to the catalog, and applies the correct classification. If it's a new item, a new identity is created on the spot, all without human intervention.

Puerto Rico’s Playbook: From Fragmented Data to Fiscal Insight

The partnership between Glass and ASG has been recognized with the inaugural Smartest AI Buy Award from the Center for Civic Futures, a distinction that highlights excellence in how governments procure and partner on AI solutions. More than an award, the project serves as a powerful case study in data-driven governance.

"With this project, we are placing a microscope over Puerto Rico’s government purchases, transforming fragmented transactions into clear, reliable, and actionable data," stated James Olmeda, Chief Information Officer at ASG Puerto Rico. He describes the result not as a technology project, but as a "foundation for smarter, more transparent, and data-driven government."

For Puerto Rico, this means having a granular, reliable view of what agencies are buying and how public funds are being used. It provides the data infrastructure needed to enforce the centralization and oversight goals of Act 73. Leaders can now see where opportunities exist to improve planning, negotiate better contracts, and ultimately deliver more value to citizens. The "living procurement intelligence ecosystem" is no longer an abstract concept; it's a functioning piece of public infrastructure.

A New Foundation for Public Accountability

The model developed in Puerto Rico is designed to be a blueprint. Given that inconsistent procurement data is a universal government problem, the potential for scalability is immense. Glass's non-disruptive, integration-focused approach is key, as it bypasses one of the biggest hurdles to government modernization: resistance to changing entrenched workflows and overhauling legacy systems.

However, the widespread adoption of such AI systems is not without its challenges. Public finance experts caution that while AI can bring incredible transparency, it also introduces new considerations. Algorithms trained on historical data could perpetuate hidden biases, and the "black box" nature of some AI can make it difficult to ensure accountability. Robust governance and human oversight are essential to ensure these tools are used ethically and do not introduce new forms of systemic risk.

By creating a single, trusted source of truth for every transaction, the Glass Master Catalog provides the bedrock for a new level of accountability. It transforms procurement data from a messy byproduct of commerce into a strategic asset for financial management. For government leaders, investors, and taxpayers who need to understand the real story behind the numbers, this invisible AI is making the flow of public money visible for the first time.