The internet,
properly synthesised.

An afternoon’s research.
Compiled in three minutes.

Unpacked doesn’t return a chat answer — it produces a complete article. The kind that would take several hours to research properly: live sources, current prices, real community opinions pulled directly from Reddit and X, independent fact-checking, and a clear verdict. Laid out like a magazine — with interactive charts, product cards, comparison tables, and an editorial hero image. You publish it under your name.

Not a product review tool. A format engine.

Unpacked classifies every question into one of nine article types and generates a completely different structure for each. A buyer’s guide gets product cards, spec tables, and a winner. An explainer gets a narrative arc. A breaking news piece gets live data and sentiment. The structure matches the question — automatically.

Five stages. Three minutes.

Every article runs through the same five-stage pipeline. A deliberately different model handles each step, which keeps errors from compounding and makes fact-checking a genuinely independent pass. Research, writing, and verification never share the same model.

The economics only just started working.

Until a year ago, the underlying model costs made a researched article unviable to generate at consumer subscription prices — loss-making from day one. The models also weren’t reliable enough to produce the structured output Unpacked depends on. Both problems are now solved.

Unlike almost every other business, Unpacked’s underlying costs fall automatically. Every new model release reduces the cost of generation with no action required. The margins only improve over time.

Why hasn’t anyone done this before?

The people with the technical capability to build this built chat interfaces — because that’s how engineers think. Nobody with editorial instincts sat at the intersection of AI pipeline design and content business models. And nobody connected it to a proven affiliate revenue model that was already generating tens of millions of dollars a year at Wirecutter.

The first-mover advantage here is real and measurable. Every article published today becomes a long-tail SEO asset that generates organic traffic for years. The content library compounds. The gap between an early entrant and a late one grows every month.

What if Wirecutter’s editorial process was automated — and anyone could use it?

Wirecutter was founded in 2011: deep research, one clear recommendation, affiliate commission on every purchase. In five years it drove $150 million in e-commerce annually. The New York Times acquired it for $30 million in 2016. By 2018 it was generating over $20 million a year for the Times. Bootstrapped throughout.

It did all of that with 60 full-time journalists, one content category, English only, with articles that took days to produce.

That’s Unpacked.

Three customers. One article. Two revenue lines on every piece.

Subscriptions from day one. Affiliate commissions the moment traffic makes them meaningful. Everything else follows a growing content library.

Built by someone who needed it.

Unpacked started as a tool built to solve a personal frustration. The moment it produced its first article, it was clear that people loved it — not just because it worked, but because it was genuinely beautiful. Then came the realisation that every article shared was bringing someone new to the platform. And that the same content driving that growth could generate serious revenue. The product, the viral loop, and the business model all emerged from the same starting point.