The Hunch Tax
Most of what your team decided this week was a guess in a data costume. For decades that was the smart move — checking cost more than the answer was worth. The cost of a trustworthy answer just collapsed, and the hunch you decided on stopped being rational.
Wrapping vs. Extracting
Most AI products are wrappers — a thin layer over a general model, restating what it already knows. That era is ending. The ones that last extract: they compute the answer no one ever wrote down, and that is the only layer that survives the model getting smarter.
Usage Is the Vanity Metric of the Intelligence Era
A tool earns product-market fit when people come back. Something sold as intelligence has to clear a harder bar — a changed decision — and most teams are still measuring the easy one.
The Question Is the Scarce Asset Now
AI collapsed the cost of the answer. The bottleneck didn't disappear — it moved to the question, the half of the workflow almost no analytics tool was built to serve.
Move the Reasoning Into the System
The one architectural choice that everything else follows from — cheap questions, an expanded market, a compounding moat, and aligned growth are not four bets, they are four consequences of moving the analytical reasoning into the system itself.
Reporting vs. Intelligence
Most products already have analytics. Almost none can be interrogated. The market split everyone gets wrong — and why the next analytics cycle won't be won on better-looking dashboards.
Curiosity Compounds
Why conversation is the only interface that aligns how a product grows with why users stay. When inquiry is the primitive, curiosity becomes the usage metric — and curiosity compounds.
The Analyst Was Never the Market
Why the demand for analysis always dwarfed the demand for analysts — and what that means for where this category grows. The analyst seat was the affordable tip of a demand curve that was always mostly underwater.
Everyone Becomes the Analyst
"Will AI replace the analyst?" is the wrong question. When the cost of asking a data question collapses, the capability spreads to the whole team — the analyst climbs to harder problems, and the new bottleneck becomes curiosity, not access.
Same Question, Different Investigation
The same question means a different investigation depending on who asks it. Why dashboards and routing systems are blind to the asker by design — and why persona-adaptive reasoning is a property of architecture, not a feature.
We're Not Their Competition. We're Their Growth Channel.
Why the defensibility question for a vertical AI company — what if the big platform builds it? — is the wrong question. Platforms profit from the growth of what lives inside them, so alignment with their gravity, not a taller wall, is the durable position.
Architecture Is the Moat, the Model Is the Engine
Why the next decade of software is decided by what the model is grounded in — not which model you wrapped. The three phases of grounding, routing versus reasoning, and why architecture outlasts the model.
From Screens to Sentences
Why intent-driven interfaces are the next strategic inflection point — moving software from navigation-driven to intention-driven.