Ideation

Forget validation. Build what no search result can show you.

It took me a decade to realise that most entrepreneurs suffocate not from a lack of capital, talent, or technology, but from breathing stale intellectual air.

Unfortunately, their people, too.

We spend our days inhaling recycled commentary, exhaling half‑formed pitches, and congratulating ourselves on the vapour in between.

We call this ritual “market research.”

The tragedy is that it feels productive while it quietly strips us of the oxygen that truly original ideas demand.

Last winter, the lesson became painfully personal. Belkins needed a fresh avenue for expansion, something outside the comfortable lanes we already dominated.

My default reflex kicked in:

• Open forty browser tabs (make my Chrome die and complain about it)

• Download industry PDFs

• Summon every Gartner quadrant ever printed.

Three evenings later, my desk was a graveyard of immaculate data pointing nowhere. Worse, my enthusiasm had drained so completely that even the prospect of a holiday felt exhausting.

Midjourney/prompt: "Imagine “Ideation”

In frustration, I shut every tab, launched OpenAI, used o3-pro and deep-search, and typed a single unfashionable question:

What would a mid‑sized US manufacturer have to believe about energy costs in 2030 to gamble on retrofitting its entire plant today?

The model returned a simulated balance sheet, an argument about US regulatory convergence, and a sketch for a sensor grid that could turn waste heat into a profit centre.

Ten minutes earlier, I was drowning in noise; now, I had a story I could not stop replaying.

Three months later, that sketch became a funded pilot, and the pilot is already expanding into a product line.

The point is not that AI handed me a billion‑dollar secret.

The point is that a well‑crafted question pierced a fog that terabytes of data could not. Ever since

I have argued that real scarcity lives in imagination, not information. AI merely raises the ceiling on what sharp questions can unlock.

Leave your questions pedestrian, and the ceiling stays invisible.

---

Why the Old Ways Collapsed

Google’s genius was flattening knowledge into an endless index (an advantage when novelty lived in what you could find).

Today, novelty lives in what you can combine.
The moment search became commoditised, every serious edge migrated one layer up, into synthesis.

Yet most accelerators still teach founders to “validate” ideas by counting competitors on a results page. They might as well teach cartography by asking you to fold a roadmap.

Two structural shifts killed that playbook:

1. Information velocity now outruns human context-switching.
   A single market signal can mutate three times between breakfast and dinner; by the time you finish a competitor matrix, half its cells are obsolete.

2. Cross-domain spill-over is the new normal.
   A tweak in Brussels’ data law can cripple a Californian ad-tech margin before lunch, while a gaming-engine update can unlock a medical-imaging breakthrough by sunset. Domains bleed into one another faster than static research models can track.

Search itself is imploding.

The latest analytics show total search-engine referrals down 15% year over year:

• Science and education sites: –10%

• Directories: –15%

• Medical resources: –31%

• Stack Overflow: –50%

• Tripadvisor: –33%

• WebMD: –50%

• Even Wikipedia: –8%

We used to Google. Now we ask Chat. And when the bot has already chewed everything up, there’s no reason to click a link.

Without that traffic, there’s no ad revenue, no community, no content. So the web retreats behind paywalls or cuts deals with AI giants.

In that climate, validation-by-Google is not just slow, it’s structurally blind.

You need an instrument that ingests multiple domains, maps them in real time, and surfaces contradictions fast enough for a human to care.

Large language models are that instrument—but only when you treat them as collaborators, not oracles.

Invisible Barrier

Bad or timid leaders recycle stale air and staff their companies with people who do the same.

The worst outcome is a venture that grows to a comfortable plateau, then stalls because neither captain nor crew ever practised synthesis. Spreadsheets masquerade as strategy; validation is mistaken for progress.

Oxygen runs out long before runway does.

---

The Question Cascade

I replaced my entire research workflow with a ritual I call the Question Cascade. It has three strokes:

1. Friction question – Identify one frustration people silently endure because the alternative seems implausible.
   Example: Factory managers write off 8% of energy as “acceptable loss.”