Same job. Different industries.
Six businesses that look nothing alike — until you look at what they actually do all day.
The matchmaker interviews two people, senses what each is looking for, and only proposes a date she believes in. Her reputation is her business.
The sales agency learns the buyer's needs and the seller's strengths — and knows a badly booked meeting costs trust in both directions.
The car-parts dealer gets a request, knows which model year takes which variant, and finds the right part among thousands of almost-right ones.
The CFO at Nordkretsen discovers the same customer in four systems under four spellings — and someone has to decide which records are actually the same company.
The harbor master knows which boats need a berth and which berths are opening up — and matches draft, beam, and waiting list every spring.
The dendrochronologist lays a sample's tree-ring series against reference collections, looking for the position where the patterns align — that's how timber gets a date.
Different industries. The same job: collect two sides, weigh them against each other, present the right pair — and stand behind it.
Why it's hard
Both sides count
It's not enough that one side wants it. A good match requires both sides' preferences to matter — otherwise it doesn't hold.
Bad pairs are expensive
A mis-hire, a wrong part, a date that never should have been booked. The cost of a bad pair always exceeds the cost of waiting for the right one.
The human doesn't scale
The judgment lives in one person who knows both sides. That works for a hundred customers. Not for ten thousand.
The ranking shortcut
The easy way is to rank: best first, list downward. Then the most popular get everything and the rest get nothing — and the list says nothing about what the other side wants. That's the popularity trap.
How we approach it
Four layers — the same method in every industry.
Signatures
Both sides described in the same language: what they are, what they seek, and what they won't compromise on.
Pair — don't rank
The matching looks for pairs where both sides are right for each other, not lists where the popular win everything.
Action — not reports
The output isn't a report to be read, but a proposal that reaches the right person, in the right channel, with the action attached. A system that acts is a colleague.
The human where the judgment sits
The final decision stays with the human — the system makes her judgment go a thousand times further.
The same pattern, built for real
eniga.ai
Law: the right answer requires the right source. eniga pairs the question with statute and precedent.
Pejl
Relationships: two partners' patterns, read together in everyday life.
TrackEid
Provenance: a sample's tree rings against the best position in the reference collections.
pinport
Boating: the right boat in the right harbour — Sweden's harbours and the boat owners looking for them. Launching soon.
The same engine under several of these examples
At the core of several of these examples is the same engine: R8 — a matching engine built on the science of stable pairs.