How it works
Matching that shows its reasoning.
Anybody can sort a list by keyword overlap and call it AI. The hard part is being right, and then being able to say why — because an attendee who does not believe a suggestion does not act on it, and an unacted-on match is worth exactly nothing to you.
The state of the art
Most “AI matchmaking” is a search box with better manners.
Tag the attendees, tag the exhibitors, count the overlaps, rank by the count. It demos well. It produces a list of people who used the same word as you, which is not the same thing as a list of people worth two hours of your day, and everybody who has attended a matchmaking event knows the difference by lunchtime on day one.
The failure is not accuracy. It is that the output cannot be argued with. A score of 82 with nothing under it tells an attendee nothing about whether to spend the afternoon on it, so they fall back on the exhibitor list and the badge colours, and the matchmaking you paid for becomes a feature nobody used.
So the engine was built to be checkable. Every match arrives as a case: what each side is here for, why these two specifically, what the risk is, and where each claim came from. People act on suggestions they can see the reasoning behind. That is the entire product thesis, and everything below follows from it.
What an attendee actually sees
One card, and no way to hide behind a number.
The score is the least interesting thing on it. What makes somebody cross a hall is the sentence underneath: a specific reason, drawn from what both of them wrote, that they can check against what they know about their own business.
Scored both ways, because a supplier who is thrilled is useless if the buyer is indifferent — and a one-way score is how matchmaking tools produce meetings that one person turns up to.
- Ranked per person, per event — never one global directory
- Scored in both directions, so it has to work for both diaries
- Every claim carries where it came from
- The caution is part of the card, not a footnote
Anna Vogel
Head of Procurement · Northwind Logistics
She is eighteen months into a retrofit programme across two chilled distribution centres and has a signed budget for Q1. You install exactly this, in this region.
Why we think so
Budget signed, decision due by Q1
— what she wrote when she registered
Two chilled sites near Rotterdam
— her company page
You ship Nordic installation partners
— your own profile
Be careful about
She cannot stop picking during a cutover. Lead with phasing, not with throughput.
An illustration of a real slate. Names and figures are made up; the shape, the sources and the caution are exactly what the product writes.
What it reads
Intent, not job titles.
A job title tells you what somebody is called. It does not tell you that they have a warehouse retrofit signed off for Q3 and eleven weeks to find somebody who can do it without stopping production. The second one is what makes a meeting worth having, and it is the thing this product asks for and everything else infers.
What they came to do
Stated per event, in their own words, and never carried between events. What somebody is buying in March is not what they are hiring for in October, and a profile that averages the two is wrong in both.
Which way they are facing
Buying, selling, sourcing, distributing, partnering, raising, investing, recruiting, looking, or here to learn. Two sellers introduced to each other is the failure mode of every directory ever built.
Who they are screening for
Industries, geographies, size. Held privately — an organiser can see what somebody offers and wants, and never the bar they set for who is worth an hour.
What the room is
Everybody registered, their companies, the stands, the sessions and the diary. A match nobody has time for is not a match, so availability is part of the answer rather than a filter applied afterwards.
It reruns whenever the room changes. Somebody registering on the morning of day two rewrites the slate for everybody they matter to, and everybody they matter to is told. An event is not a fixed list; treating it as one is why so much matchmaking is stale by the time the doors open.
What it refuses
The rules it will not break for a better-looking number.
It would be easy to fill a slate. Filling it with meetings people actually take is a different job, and most of that job is knowing what not to suggest.
It will not introduce two people who came together
Colleagues from the same stand are not a networking opportunity. Neither are two people who already have a meeting, or who have already met and said it was not useful.
It will not put a hidden person on a slate
Somebody who has made themselves invisible to a company or a person is absent from the slate, silently, with no mark on the other side — and that holds even when the person running the event is the one they hid from.
It will not claim what it cannot show
Every statement in a match carries the source it was drawn from. Anything a person wrote about themselves outranks anything written about them, and they can correct or replace any of it.
It will not pretend a meeting is perfect
Where there is a reason to be careful — a mismatch in scale, a wrong region, a timing problem — it says so in the same card. A tool that only ever enthuses is one nobody believes twice.
End to end
Registration to report, in one system.
The matching is the reason people buy it. What makes it work on the day is that the same system holds the tickets, the door, the agenda and the diary — so the engine knows who actually turned up, and the report knows what the engine did.
1 · They register
Through your page, your ticket, or an import of the list you already hold. Fifteen hundred rows with the progress on screen, resumable, and cancellable without leaving half a list behind.
2 · They say what they came for
A short, specific question rather than a profile form: what are you offering, what are you looking for, who is worth your time. This is the input the whole engine runs on, and getting good answers is a craft — there is a guide on it.
3 · The slate is written
Ranked for that person at that event, with the case for each one in plain words. Not a directory to search. A list to act on, waiting when they open the app.
4 · They ask for the time
A meeting request with the reason attached, answered in one tap. Where you want oversight, requests can route through your team before they land.
5 · The diary sorts itself
Sessions they are attending block their time automatically. Nobody is offered a slot they are already in a workshop for, and the suggestion is a real free slot for both people.
6 · They walk in prepared
A short brief before the meeting: who the other person is, what they came for, and what to open with. Written for one side and never visible to the other, and never to you.
7 · You get the proof
Meetings agreed, meetings held, who turned up and what came of it — per exhibitor, per sponsor, per stand. The document you take into next year's budget conversation.
Each of those is described properly on the platform page.
What we do not claim
It is a considered suggestion, not a guarantee.
We do not promise every match is right. We promise every match tells you why it was made, so you can decide in ten seconds instead of finding out in thirty minutes. Over nine hundred people that difference is the entire event.
We also do not promise anybody turns up. We can tell you who did not, which is a different and considerably more useful thing to have in front of you when you are deciding who to invite next year.
See it against your own event.
The fastest way to judge any of this is to point it at a real room. Tell us what you run, roughly how many come, and what you wish had happened last time.