Context is the new data

"Data is the new oil" defined the strategic thinking of the 2010s. Companies invested billions in data infrastructure: collection, storage, transformation, and analysis. The data warehouse became standard infrastructure. The companies with the best data operations (Google, Amazon, Netflix, Uber) built competitive advantages that proved durable.

The data advantage is now table stakes. Every company has a data warehouse. Every company runs analytics. Every company has dashboards. The infrastructure is commoditised, and while the quality of analysis still varies, the fundamental capability of collecting and querying structured data is no longer a differentiator.

The next layer of competitive advantage is context: the unstructured, qualitative knowledge that explains what the data means and what to do about it.

Data tells you what. Context tells you why.

Your data shows that enterprise churn increased 15% last quarter. Your context, if you could access it, would tell you that three of the churned accounts cited the same feature gap, that the product team deprioritised that feature six months ago based on reasoning that no longer applies, that the sales team has been hearing about the gap from prospects for months, and that a competitor launched exactly that feature in their last release.

The data is the signal. The context is the intelligence. A company that has both makes better decisions, faster, than a company that has only the data.

The reason context has historically been less valuable than data is that context had no infrastructure. You couldn't warehouse it, query it, or pipe it to analytical tools. It lived in people's heads, in Slack threads that scrolled away, in meetings that were never recorded. The context existed but it was operationally inaccessible, which meant it couldn't be treated as a strategic asset.

That's changing.

The context infrastructure layer

Self-writing documentation, semantic search, and open AI protocols have made it possible, for the first time, to build context infrastructure that parallels data infrastructure.

Collection: Connections to the tools where context is created (Slack, GitHub, meetings, email, Google Drive) ingest the raw context continuously.

Transformation: Self-writing docs convert raw activity into structured, citable documentation. The Slack discussion becomes a decision record. The meeting becomes a searchable transcript with extracted insights. The PR becomes a system documentation update.

Query: Semantic search and MCP make the context queryable by humans and AI agents, in the same way SQL and BI tools make data queryable.

This infrastructure layer, the context warehouse, is what allows context to be treated as a strategic asset for the first time. The knowledge that was previously locked in ephemeral channels and people's memories becomes persistent, searchable, and compounding.

The compounding advantage

Context infrastructure has a property that data infrastructure largely doesn't: it compounds in value over time in a way that accelerates rather than plateaus.

A data warehouse with six months of transaction data is useful. A data warehouse with five years of transaction data is more useful, but the marginal value of each additional month diminishes as the historical patterns become well-established.

A context warehouse with six months of accumulated decisions, discussions, and institutional knowledge is useful. A context warehouse with five years of accumulated context is dramatically more useful, because the connections between older context and newer context produce insights that neither alone could generate. The product decision from two years ago explains the architectural constraint that's causing the engineering bottleneck today. The client conversation from eighteen months ago predicted the competitive threat that's materialising now. The research from last year answers the question that this year's strategy depends on.

The companies that start building context infrastructure now will have accumulated context that can't be fast-forwarded or bought. A competitor who starts in two years begins with zero accumulated context, regardless of how much they spend on the infrastructure itself. The context is the moat, and the moat deepens with time.

Who moves first

Data warehouse adoption followed a predictable pattern: technically sophisticated companies built them first (late 2000s), mainstream adoption followed (mid 2010s), and by the early 2020s they were universal. The companies that moved first had five to ten years of accumulated data advantage over late adopters.

Context warehouse adoption is at the beginning of the same curve. The companies investing in context infrastructure now, connecting their tools, enabling self-writing documentation, building the query layer, are the ones that will have years of accumulated context when the mainstream catches up.

The infrastructure is available. The question is which companies recognise the parallel to data warehousing early enough to build the advantage.

Frequently asked questions

Isn't this just knowledge management by a different name? Knowledge management has historically meant wikis, documents, and manual curation. A context warehouse is infrastructure: automated collection, transformation, storage, and query. The relationship is roughly analogous to "filing cabinet" versus "data warehouse." Both store information. The infrastructure version is categorically more powerful.

How does this relate to AI strategy? A context warehouse is AI infrastructure. Every AI tool deployed in the enterprise benefits from richer context, and the context warehouse is what provides it. Companies that build context infrastructure before deploying AI tools get dramatically more value from their AI investment than companies that deploy AI into a context vacuum.

Is this only relevant for large companies? The pattern scales down. A 20-person startup that builds context infrastructure from day one will have richer institutional memory at 100 people than a company that starts building at 100. The earlier you start, the more context accumulates, and the more valuable the infrastructure becomes.

Related reading: What is a context warehouse?, Your data warehouse handles the numbers, The memory is the moat, Your company isn't ready for AI. Related pages: Self-writing docs, Connections, MCP.