Oklahoma Firm Debuts AI to Predict Public Opinion with County-Level Precision

OKLAHOMA CITY, OK – May 19, 2026 – In a move that could reshape how organizations communicate with the public, Oklahoma-based consultancy Saxum has launched Foundry DST, an intelligence platform that claims to predict how a message will be received before it ever goes public. The new tool creates a new category of "anticipatory sentiment," shifting communications strategy from a reactive, backward-looking discipline to a proactive, predictive science.

Foundry DST is designed to function as a decision support tool, providing a granular, county-by-county analysis of likely public sentiment for any piece of content—be it a policy announcement, a marketing campaign, or a crisis response statement. This allows organizations to test, refine, and optimize their messaging in a simulated environment, grounding consequential decisions in data rather than intuition alone.

From Post-Mortem to Pre-Mortem in Communications

For decades, the standard communications playbook has been to launch a campaign and then measure its impact through post-launch tools like polls, focus groups, and media monitoring. This reactive approach often means insights arrive too late to influence strategy. Foundry DST aims to invert this model entirely.

"The organizations that will lead in communications over the next decade are the ones that know more before they move," said Hart Brown, President, AI & Transformation at Saxum, the firm behind the new platform. "The tools that exist today were built to measure the past. Foundry DST is built for the moment when strategy can still change, when risk can still be reduced and decisions can still be strengthened. This tool fundamentally changes the way teams think about messaging, and is the benchmark of how to operate moving forward."

The platform works by ingesting a message—text, image, or even video—and returning a simple, color-coded classification for every one of Oklahoma's 77 counties: Green for likely positive reception, Yellow for mixed or neutral, and Red for likely resistance. Each classification is accompanied by a confidence score and a narrative explaining the reasoning, drawn from a deep well of local data.

When the system flags a county with a Yellow or Red sentiment, it doesn't just diagnose the problem. It becomes prescriptive, identifying specific message frames that are more likely to resonate, flagging words or phrases to avoid, and even delivering optimized rewrites. This iterative process stands in stark contrast to traditional research methods, which can cost tens of thousands of dollars and take up to 16 weeks to field for a single, static result.

The Power of Hyper-Local Intelligence

While many AI-driven platforms like Brandwatch and Sprinklr offer predictive analytics for broad market trends, Foundry DST's unique advantage lies in its hyper-local, sub-state precision. The engine driving this capability is a proprietary dataset called the "Oklahoma FactBook."

This is not simply scraped census data. The FactBook is built on what Saxum describes as over two decades of its cross-sector work in the state, encompassing public health, energy, government affairs, and tribal partnerships. It contains more than 24 distinct data dimensions for each county, but its most unique feature is the integration of intelligence shaped by Oklahoma’s 39 federally recognized tribes.

This deep, localized data architecture is what makes the platform’s intelligence defensible and difficult for national competitors to replicate. In a state where geography, culture, trust dynamics, and tribal sovereignty materially shape how messages land, off-the-shelf national datasets often miss critical nuances. Rebuilding the quantitative data from public sources would be a years-long endeavor; rebuilding the qualitative relationships and trust behind the tribal insights would take far longer. This blend of quantitative data and qualitative, relationship-based intelligence creates a formidable barrier to entry and underscores the platform's claim to understand the state on a deeper level.

Building a More Reliable Crystal Ball

At a time of widespread concern over AI "hallucinations"—where models generate plausible but false information—Saxum has built Foundry DST with a multi-layered architecture designed to ensure reliability and reduce risk. The system doesn't rely on a single large language model. Instead, it runs four independent frontier AI models simultaneously.

The outputs from these four models are then statistically normalized into a single, unified view. This design serves a dual purpose: it averages out the idiosyncrasies and potential biases of any single model, and it acts as an early warning system. When there is high variance between the models' conclusions, the output is automatically flagged for review by a human analyst before being delivered to the client.

This "human-in-the-loop" approach, combined with the multi-model design, aligns with industry best practices for deploying AI in high-stakes applications. It acknowledges the limitations of current AI technology while leveraging its power, creating a system that is both fast and accountable. The result is an anticipatory analysis that is not just a black-box prediction, but a confidence-rated insight mediated by expert human judgment.

A Strategic Advantage Born in the Heartland

The development of such a sophisticated tool in Oklahoma City, rather than a traditional tech hub, highlights a growing trend of regional innovation. Saxum has leveraged its 20-year history and deep roots in the state to solve a problem that plagues organizations everywhere.

"The tool is built in Oklahoma for Oklahomans, but the problem it solves is not unique to the state," Brown noted in the announcement. "It is the fundamental gap in how communications decisions get made everywhere: organizations invest in research that is either too slow to inform strategy, too broad to be actionable, or too backward-looking to matter when the decision is still on the table."

The potential applications are extensive. Political campaigns could use the tool to micro-target messaging in swing counties. State public health agencies could test the reception of vaccination campaigns to overcome localized pockets of hesitancy. Energy companies planning major infrastructure projects could anticipate community concerns and refine their outreach strategies to build trust more effectively. For any organization navigating the complex patchwork of cultures and opinions across a state, the ability to conduct a "pre-mortem" on public opinion offers a significant structural advantage, potentially saving millions in misdirected campaigns and protecting reputations from avoidable missteps.

While initially focused on Oklahoma, the underlying model—combining advanced AI with deep, proprietary regional data and expert human oversight—provides a blueprint for how communications intelligence could evolve in other complex environments around the world.