Every enterprise leader has heard the promise by now. AI will transform decision-making, unlock productivity, and reshape competitive advantage. Few would disagree. Yet the lived reality inside most organizations looks different, a proliferation of pilots, a graveyard of proof of concepts, and a persistent question that keeps surfacing in boardrooms. Why is it not scaling?
In the latest episode of Unfiltered Stories | HFS & MathCo: AI Models Are a Utility, Context Is Your Identity, Anuj Krishna, Cofounder and President – Technology and Growth at MathCo, joined HFS Research President Saurabh Gupta to share crucial insights and offered a pointed answer. The bottleneck is not the model, it is context, the accumulated knowledge of how an organization actually works, the decisions it makes, and the reasoning behind them. According to a survey conducted by HFS Research in collaboration with MathCo, nearly two-thirds of enterprises admit that they are not aware where this knowledge resides. It sits scattered across documents, unwritten processes, and the heads of a few tenured employees. Models can now be commoditized almost overnight, Anuj Krishna adds, but institutional knowledge cannot.
The Uncomfortable Truth About “Context”
Context has become one of 2026’s most overused buzzwords, invoked by hyperscalers, platform vendors, and systems integrators alike, each claiming a stake in “owning” it. But as Anuj Krishna notes, context is not a data problem alone. It is broader and more dynamic. What questions the business needs answered, what decisions get made, and what happens after those decisions are made. That is a different mandate than the traditional semantic layer, and it explains why context governance has no natural owner today. It does not sit neatly with IT, with the business, or with a single center of excellence, which is precisely why it keeps falling through the cracks.
Anuj’s analogy is worth sitting with. Context is like a photograph of a river. Accurate the moment it is taken, obsolete the moment after. Treating context as a static asset, a one-time mapping exercise, guarantees it decays. What enterprises need instead is a living system, with clear stewardship, continuous feedback loops, and a deliberate effort to keep “verification cost” from spiraling as more agents are deployed to act on that knowledge.
From Fragmented AI to Compounding Intelligence
This is where Thinking Architecture comes in. Rather than treating copilots, forecasting tools, and agents as isolated point solutions, it is built on a simple but consequential premise that intelligence should compound and not fragment. High accuracy with low adoption is not success, it is a symptom of AI that never connects to the rest of the enterprise. MathCo’s Thinking Architecture brings context, decision-making, and agentic execution into one coherent system, so that knowledge captured in one part of the business strengthens judgment everywhere else.
Critically, this is not an argument for outsourcing your intelligence to a platform. The organizations that will pull ahead are the ones that treat their context, their tribal knowledge, their decision logic, and their operating “personality” as a proprietary asset. It can be built with the ecosystem, but it must be owned outright. Rent the tooling. Own the knowledge.
Where To Start
For leaders wondering how to make this tangible, three starting points have proven effective:
- Audit before you architect. Most organizations already have more structured knowledge than they realize. It is buried in documentation, decks, and platforms. The first move is surfacing and organizing it, not starting from a blank page.
- Pick your entry point deliberately. Some enterprises launch a full context program, others start with a single function and build outward, others begin with a horizontal knowledge corpus. There’s no single right answer, but the system must be designed to expand.
- Assign ownership now. Context without a steward decays. Someone, a function, a role, or a program has to be accountable for keeping it current and correct.
The Strategic Takeaway for Leaders
Models will become a utility. Building agents will become table stakes. Neither will differentiate one enterprise from another for long. What will set companies apart is context, how it is captured, governed, and continuously renewed as a living asset. The organizations that start building that discipline now will spend the next decade compounding an advantage that competitors cannot simply license or replicate.
Explore the full conversation here: Unfiltered Stories | HFS & MathCo: AI Models Are a Utility, Context Is Your Identity – HFS Research
For a deeper dive into the data, read the associated Take 5 Report “Fund the Context Layer and Close the 2.5x AI Ambition Gap”
