How it works

The whole system, no black box.

Most matching products keep the mechanics hidden. Piaar doesn't. Here's exactly how the platform finds people for you, scores them, and decides what to show.

01

Deep onboarding

The first thing Piaar asks isn't your job title. It's your skills — specifically, three to ten you can teach or help with, and three to ten you'd love to learn or hire for. These are the two sides of every profile.

Skills are pulled from a curated catalog that's growing with the user base. If yours isn't there, you can propose it — a small flag marks it as "custom" while we review it. Custom skills don't count toward matching until they're promoted into the catalog, usually within a week.

Why both sides matter

Almost every other platform asks you what you do. Piaar asks what you have AND what you need. That second axis is what makes mutual fit possible — and mutual fit is what makes the conversation actually happen.

02

Semantic matching

When your profile is set, the matching service runs a query: your "looking-for" skills against everyone else's "offer" skills, and your "offer" skills against everyone else's "looking-for". It's bidirectional by design.

The match is semantic, not keyword. "Machine Learning" finds people who wrote "Deep Learning", "AI Engineering", or "Neural Networks". You wouldn't expect those to be different matches as a user, and the system doesn't treat them as different either.

What "semantic" actually means here

Each skill is embedded as a vector in a model trained on professional language. Two skills are considered neighbors if their vectors are close. We can show you which neighbors contributed to a given match score, on request — that's the EU AI Act commitment, not a feature flag.

03

Match scores

Every match comes with an integer score from 0 to 100. The score reflects mutual fit — how well your offers match their looking-for, AND how well their offers match your looking-for. Single-sided matches don't score high, even if the overlap on one side is perfect.

Scores below 40 are filtered out before they reach you. You won't see "12% match" cards. If we don't have a real reason to surface someone, we don't.

87
Strong match

80+ · Teal ring

64
Good match

60–79 · Teal mid

48
Weak match

40–59 · Grey

Example breakdown — a score of 87

Two people, James (offers UX research, seeks Flutter) and Sarah (offers Flutter, seeks UX research). Same city, both within radius. The score gets built like this:

Skill overlap (mutual)Both sides fully covered
95
Skill specificity"Flutter" beats "mobile dev"
82
Distance penalty4.2 km apart
90
Recency bonusBoth profiles updated this month
80
04

Three views, one query

Home, Discover-list, and Discover-map are three different ways of looking at the same underlying match results. They don't run separate queries. Change the radius in Discover, the Home digest refreshes on next return.

  • Home — your personalized digest. Top three to five matches and two to three nearby events. Read-mostly. No filters.
  • Discover list — every match in scrollable, sortable form. Filters exposed for radius, categories, and sort order.
  • Discover map — every match as a pin on an OpenStreetMap. Tap a pin, read the peek card, tap through to a full profile.
05

Graduated connection

Piaar deliberately isn't a "match" or "swipe" product. The connection model is a ladder:

  • View profile — anyone with a discoverable profile is viewable. Viewing isn't contacting.
  • Send a message request — opens a one-off thread. The recipient can accept (it becomes a real chat), decline (you're told), or ignore (it expires in 14 days).
  • Send a connection request — invites them into your Trusted Network. Accepted connections appear in both Network tabs and unlock easier ongoing messaging.

Message is the lighter ask. Connect is the bigger one. The product makes the hierarchy clear so you don't accidentally do the wrong one.

06

Privacy controls, always on

Two binary controls are live now, with more coming. Both default on, both reachable from one screen in Account settings.

Discoverable profile

When off, you don't appear in any other user's match results, list, or map. You can still browse others. This is enforced at the query layer — not the UI layer — so a non-discoverable user is never accidentally rendered to anyone else.

Precise location

When off, your location is obfuscated to a ~1 km grid square for map display. The matching service still uses your real location for radius math (otherwise distance wouldn't make sense), but pins snap to grid centers, and the profile shows "around 3 km" instead of "2.7 km".

See it in action.

Set up takes under three minutes. Matches arrive the same day.

Download on theApp Store Get it onGoogle Play