The Missing Link: How One Startup Is Teaching AI to Speak Business

BOULDER, Colo. – July 28, 2026 – In the relentless race to integrate artificial intelligence into every facet of corporate life, a quiet crisis is unfolding. It’s not a crisis of computing power or algorithmic sophistication, but one of trust. Enterprises are discovering that even the most advanced AI agents can be confidently, catastrophically wrong when interpreting the lifeblood of the organization: its structured data. Addressing this foundational challenge, a Boulder-based startup, Credible Data, today announced a $10 million seed round to scale its platform, designed to serve as a universal translator between complex business logic and AI.

The funding, led by prominent venture firms Gradient, SignalFire, and K5 Global, with participation from luminaries like pandas creator Wes McKinney, signals a growing recognition that AI’s true potential is shackled by its inability to grasp business context. Without a consistent understanding of what data actually means, the promise of AI-driven decision-making remains fraught with risk.

A Crisis of Context

For years, enterprises have poured resources into data warehouses like Snowflake, Google BigQuery, and Postgres, creating vast, structured repositories of every transaction, customer interaction, and operational metric. While AI has made strides in processing unstructured information from documents and knowledge bases, it often falters when faced with these databases. The reason is simple yet profound: a database table is not a business.

An AI agent sees rows and columns; it does not inherently understand that a company’s definition of "monthly recurring revenue" might exclude certain one-time fees, or that an "active user" is defined differently by the marketing and product teams. This semantic gap is where trust breaks down. A recent industry survey found that 57% of enterprises have traced confidently wrong AI answers directly back to this missing or inconsistent business context. The result is a landscape of conflicting dashboards, unreliable AI-generated reports, and a deep-seated executive skepticism that undermines billion-dollar technology investments.

"Every organization attaches its own meaning to its data -- even a term as familiar as 'revenue' can mean completely different things from one company to the next," said Credible Data CEO and founder Kyle Nesbit. "AI can't be trusted with enterprise data until it understands that meaning. The winners in enterprise AI will be the ones who deliver their data's meaning as context wherever decisions get made, without locking themselves into a single vendor's stack."

An Open-Source Foundation for Trust

Credible Data’s answer to this problem is not another proprietary black box. Instead, its platform is built upon Malloy, an open-source semantic modeling language developed by Lloyd Tabb, the visionary founder of the pioneering business intelligence tool Looker. Malloy functions as a “headless” semantic layer—a central, governable source of truth that defines all of an organization's key metrics, entities, and relationships as code.

This approach decouples business logic from any single application. The Malloy model acts as a universal Rosetta Stone, providing consistent, executable context to any system that needs it, from an AI agent generating a market analysis to a traditional BI dashboard tracking quarterly performance. By building on an open-source standard, Credible Data offers enterprises a path away from the vendor lock-in that has long plagued the data industry. The models are portable, versionable in Git, and transparent.

"Existing data warehouse and pipeline solutions were never built for AI," noted Zach Bratun-Glennon, General Partner at Gradient, Google's AI-focused venture fund. "As enterprises rush to deploy AI... they are finding their ability to trust this new technology undermined by conflicting outputs... Credible Data has built a platform that solves this challenge with unprecedented agility and flexibility."

While Malloy provides the powerful, expressive core, Credible Data wraps it in the enterprise-grade armor required for mission-critical workloads: robust security, role-based access controls, governance workflows, and detailed auditability. It transforms an open-source language into a scalable engine for trustworthy AI.

The Human Element: A Pedigree of Data and AI

Technology alone is rarely the whole story. The credibility of the solution is deeply intertwined with the experience of the team building it. Here, Credible Data presents a compelling narrative. CEO Kyle Nesbit previously led Business Intelligence and Data Analytics at Google Cloud, where he was responsible for spearheading the integration of Looker—the very technology whose philosophical successor, Malloy, now powers his company. His career has given him a front-row seat to the persistent gap between data infrastructure and reliable analysis.

Complementing this data-centric expertise is Head of Product James Swirhun, who spent eight years at Google working on flagship AI/ML products, including the Gemini family of models. His intimate understanding of what modern AI agents require to function effectively provides the crucial other half of the equation. This fusion of deep data systems knowledge with cutting-edge AI product experience positions the company to bridge the chasm that has stymied so many others.

From Experiment to Production

For early customers, the impact is already moving from theoretical to tangible. Peter Nummerdor, SVP of Product for advertising technology firm VideoAmp, is using the platform to ground its AI systems in a shared, governed understanding of its complex business metrics. The insights their systems provide drive decisions for some of the largest players in the advertising industry, leaving no room for error or inconsistency.

"Credible gives our AI systems the same governed understanding of our metrics that our best analysts carry in their heads," Nummerdor stated. "That's what moved us from experimenting with AI to putting it in front of customers with confidence."

This statement captures the essence of the transformation at hand. By codifying and scaling the institutional knowledge previously held only by a few key experts, Credible Data’s platform allows organizations to de-risk their AI initiatives. It provides a systematic way to ensure that as AI becomes more autonomous, it remains anchored in the verifiable, nuanced reality of the business it is meant to serve.