The Missing Pillar in Agentic AI: Why I Joined AtScale

Recently, while debating my next career chapter with a close friend, they highlighted something I had almost begun to take for granted: the most important work our civilization is doing right now is tied to the advancement of AI. It will act as a massive accelerant, driving breakthroughs across every industry and giving us a chance to solve global challenges previously viewed as out of reach.

For the past decade, I’ve had a front-row seat to this revolution. During my tenure at DataRobot, I transitioned from Ph.D. Engineer to a 9-figure P&L owner, eventually leading a 300+ person global organization focused on driving ruthlessly practical ROI from AI. I personally led our strategic evolution from predictive machine learning (ML) to Agentic AI, a shift that wasn’t just a technical milestone but a commercial accelerator that grew our new bookings by 51% YoY.

But through that journey, I identified the primary failure mode of modern AI: The Technical Trust Gap.

From “Ask Mode” to “Agent Mode” — And the Wall We Keep Hitting

As we move from “Ask Mode” (chatbots) to “Agent Mode” (autonomous systems), we are hitting a massive technical wall. The core challenge isn’t the model. It’s Context Drift.

As the number of tokens in an agent’s context window grows, the agent begins to lose focus. I call this Context Rot. When you feed an agent raw, noisy tables and hundreds of schema definitions, you aren’t giving it more information. You’re giving it more opportunities to fail.

To combat this, advanced systems need Selective Retrieval: pulling in only the specific business logic that matters, precisely when it matters. This isn’t just good engineering practice. It’s the architectural requirement that separates AI agents that are impressive in a demo from AI agents that are trusted in production.

This is exactly why I joined AtScale.

A universal semantic layer is the ultimate strategy for Context Reduction. It allows an agent to query a single, compressed business definition that is accurate and deterministic, rather than traversing a labyrinth of raw, noisy data. AtScale has spent over a decade building exactly that foundation, and no one in the market has gone deeper on the problem of governed, enterprise-scale semantics.

The Three Pillars of Enterprise AI

To move beyond experimentation, AI agents require three distinct pillars to operate reliably:

1. Reasoning (LLMs): The cognitive engine that processes intent.
2. External Context (The Web): Tools that ground the model in real-time market reality.
3. Internal Context (Governed Semantic Layer): The single source of truth that ensures agents understand proprietary data accurately, consistently, and with full auditability.

The industry has poured billions into the first two pillars. Internal Context is the missing foundation.

Right now, organizations are trying to solve this with ad-hoc manual solutions: the equivalent of hard-coding business logic directly into a prompt. That approach doesn’t scale, it doesn’t govern, doesn’t work for multi-agent, and barely accommodates multiuser. You cannot have autonomous agents making real business decisions on data they don’t truly understand.

AtScale provides that Internal Context layer. It’s the governed semantic foundation that gives AI agents access to business logic defined once, version-controlled, and enforced everywhere — across agents, dashboards, applications, and analytical workflows.

Deterministic AI vs. Probabilistic Guesswork

Here’s the distinction that matters most as we move into the agentic era.

Summarizing a document can have many acceptable answers. Exact calculations have one answer.

Predictive AI could tolerate approximation. If a model was 85% accurate, that was a win as it improved a prediction of a future outcome beyond what a person could typically achieve. Agentic AI cannot. When an autonomous system triggers pricing changes, inventory allocations, or financial decisions, the logic must be exact. In AtScale’s testing against TPC-DS benchmarks, LLMs working directly with raw data are incorrect more than 80% of the time. They don’t just return wrong numbers. They fabricate relationships that don’t exist.

If an AI agent calculates Gross Dollar Retention differently than your CFO, ignoring time shifts, business caveats, or metric definitions, that isn’t a hallucination. It’s a tangible danger to your business.

AtScale eliminates that risk. By routing every AI query through governed business definitions, the semantic layer ensures that “Product Usage” means the same thing to the LLM as it does to your engineering team, your finance team, and your CEO.

Why AtScale Is Required Infrastructure

Throughout my career, I’ve viewed AI governance not as a compliance hurdle but as a commercial differentiator. Organizations that get governance right move faster, because their teams trust the outputs.

AtScale provides the Internal Context that acts as the USB-C of enterprise AI: a standardized, machine-readable interface for business logic. Through open standards like the Model Context Protocol (MCP) and the Semantic Modeling Language (SML), AtScale allows AI agents to access precise, governed definitions that match exactly how the finance team calculates those same metrics.

My career has been defined by aligning technical depth with commercial execution. AtScale’s mission is the technological equivalent: getting AI systems and business leaders to speak the exact same language, and making sure that conversation scales across every tool, every cloud, and every agent in the enterprise.

The Road Ahead

We are moving from early adoption to operational standard. The organizations investing in semantic infrastructure now will be the ones that can trust their AI systems to make autonomous decisions tomorrow.

Leadership is ultimately about stewardship. Building trusted AI is the ultimate form of enterprise stewardship, over your data, your decisions, and the people who depend on both.

I’m incredibly excited to join Chris Lynch and the team at AtScale to lead global go-to-market, operations, and AI innovation. The foundation is built. The market is ready. 

Let’s get to work.

AtScale is hiring, check out our careers page for current openings. To learn more about AtScale, request a demo.